bridge_driver.c 175.1 KB
Newer Older
1
/*
2
 * bridge_driver.c: core driver methods for managing network
3
 *
4
 * Copyright (C) 2006-2016 Red Hat, Inc.
5 6 7 8 9 10 11 12 13 14 15 16 17
 * Copyright (C) 2006 Daniel P. Berrange
 *
 * This library is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * This library is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
18
 * License along with this library.  If not, see
O
Osier Yang 已提交
19
 * <http://www.gnu.org/licenses/>.
20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41
 *
 * Author: Daniel P. Berrange <berrange@redhat.com>
 */

#include <config.h>

#include <sys/types.h>
#include <sys/poll.h>
#include <limits.h>
#include <string.h>
#include <stdio.h>
#include <stdarg.h>
#include <stdlib.h>
#include <unistd.h>
#include <errno.h>
#include <sys/utsname.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <signal.h>
#include <pwd.h>
#include <sys/wait.h>
#include <sys/ioctl.h>
42
#include <net/if.h>
43
#include <dirent.h>
44 45 46
#if HAVE_SYS_SYSCTL_H
# include <sys/sysctl.h>
#endif
47

48
#include "virerror.h"
49
#include "datatypes.h"
50
#include "bridge_driver.h"
51
#include "bridge_driver_platform.h"
52
#include "network_conf.h"
53
#include "device_conf.h"
54
#include "driver.h"
55
#include "virbuffer.h"
56
#include "virpidfile.h"
57
#include "vircommand.h"
58
#include "viralloc.h"
59
#include "viruuid.h"
60
#include "viriptables.h"
61
#include "virlog.h"
62
#include "virdnsmasq.h"
63
#include "configmake.h"
64
#include "virnetdev.h"
65
#include "virpci.h"
66 67
#include "virnetdevbridge.h"
#include "virnetdevtap.h"
68
#include "virnetdevvportprofile.h"
69
#include "virdbus.h"
70
#include "virfile.h"
71
#include "virstring.h"
72
#include "viraccessapicheck.h"
73
#include "network_event.h"
74
#include "virhook.h"
75
#include "virjson.h"
76

77
#define VIR_FROM_THIS VIR_FROM_NETWORK
78
#define MAX_BRIDGE_ID 256
79

80 81 82 83 84 85 86
/**
 * VIR_NETWORK_DHCP_LEASE_FILE_SIZE_MAX:
 *
 * Macro providing the upper limit on the size of leases file
 */
#define VIR_NETWORK_DHCP_LEASE_FILE_SIZE_MAX (32 * 1024 * 1024)

87 88
VIR_LOG_INIT("network.bridge_driver");

89
static virNetworkDriverStatePtr network_driver;
90

91 92 93 94 95 96 97 98
static virNetworkDriverStatePtr
networkGetDriver(void)
{
    /* Maybe one day we can store @network_driver in the
     * connection object, but until then, it's just a global
     * variable which is returned. */
    return network_driver;
}
99

100
static void networkDriverLock(virNetworkDriverStatePtr driver)
101
{
102
    virMutexLock(&driver->lock);
103
}
104
static void networkDriverUnlock(virNetworkDriverStatePtr driver)
105
{
106
    virMutexUnlock(&driver->lock);
107 108
}

109 110 111 112
static dnsmasqCapsPtr
networkGetDnsmasqCaps(virNetworkDriverStatePtr driver)
{
    dnsmasqCapsPtr ret;
113
    networkDriverLock(driver);
114
    ret = virObjectRef(driver->dnsmasqCaps);
115
    networkDriverUnlock(driver);
116 117 118 119 120 121 122 123 124 125 126
    return ret;
}

static int
networkDnsmasqCapsRefresh(virNetworkDriverStatePtr driver)
{
    dnsmasqCapsPtr caps;

    if (!(caps = dnsmasqCapsNewFromBinary(DNSMASQ)))
        return -1;

127
    networkDriverLock(driver);
128 129
    virObjectUnref(driver->dnsmasqCaps);
    driver->dnsmasqCaps = caps;
130
    networkDriverUnlock(driver);
131 132 133
    return 0;
}

134
static int networkStateCleanup(void);
135

136 137
static int networkStartNetwork(virNetworkDriverStatePtr driver,
                               virNetworkObjPtr network);
138

139 140
static int networkShutdownNetwork(virNetworkDriverStatePtr driver,
                                  virNetworkObjPtr network);
141

142 143
static int networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
                                      virNetworkObjPtr network);
144

145 146
static int networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
                                         virNetworkObjPtr network);
147

148
static int networkStartNetworkExternal(virNetworkObjPtr network);
149

150
static int networkShutdownNetworkExternal(virNetworkObjPtr network);
151

152 153
static void networkReloadFirewallRules(virNetworkDriverStatePtr driver);
static void networkRefreshDaemons(virNetworkDriverStatePtr driver);
154

155 156 157 158 159
static int networkPlugBandwidth(virNetworkObjPtr net,
                                virDomainNetDefPtr iface);
static int networkUnplugBandwidth(virNetworkObjPtr net,
                                  virDomainNetDefPtr iface);

160
static void networkNetworkObjTaint(virNetworkObjPtr net,
161
                                   virNetworkTaintFlags taint);
162

163 164 165
static virNetworkObjPtr
networkObjFromNetwork(virNetworkPtr net)
{
166
    virNetworkDriverStatePtr driver = networkGetDriver();
167 168 169
    virNetworkObjPtr network;
    char uuidstr[VIR_UUID_STRING_BUFLEN];

170
    network = virNetworkObjFindByUUID(driver->networks, net->uuid);
171 172 173 174 175 176 177 178 179 180
    if (!network) {
        virUUIDFormat(net->uuid, uuidstr);
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching uuid '%s' (%s)"),
                       uuidstr, net->name);
    }

    return network;
}

181 182 183
static int
networkRunHook(virNetworkObjPtr network,
               virDomainDefPtr dom,
184
               virDomainNetDefPtr iface,
185 186 187 188 189 190 191 192 193
               int op,
               int sub_op)
{
    virBuffer buf = VIR_BUFFER_INITIALIZER;
    char *xml = NULL, *net_xml = NULL, *dom_xml = NULL;
    int hookret;
    int ret = -1;

    if (virHookPresent(VIR_HOOK_DRIVER_NETWORK)) {
194 195 196 197 198 199
        if (!network) {
            VIR_DEBUG("Not running hook as @network is NULL");
            ret = 0;
            goto cleanup;
        }

200 201
        virBufferAddLit(&buf, "<hookData>\n");
        virBufferAdjustIndent(&buf, 2);
202
        if (iface && virDomainNetDefFormat(&buf, iface, NULL, 0) < 0)
203
            goto cleanup;
204 205
        if (virNetworkDefFormatBuf(&buf, network->def, 0) < 0)
            goto cleanup;
206
        if (dom && virDomainDefFormatInternal(dom, NULL, 0, &buf) < 0)
207 208 209 210 211
            goto cleanup;

        virBufferAdjustIndent(&buf, -2);
        virBufferAddLit(&buf, "</hookData>");

212
        if (virBufferCheckError(&buf) < 0)
213 214
            goto cleanup;

215
        xml = virBufferContentAndReset(&buf);
216 217 218 219 220 221 222 223
        hookret = virHookCall(VIR_HOOK_DRIVER_NETWORK, network->def->name,
                              op, sub_op, NULL, xml, NULL);

        /*
         * If the script raised an error, pass it to the callee.
         */
        if (hookret < 0)
            goto cleanup;
224 225

        networkNetworkObjTaint(network, VIR_NETWORK_TAINT_HOOK);
226 227 228
    }

    ret = 0;
229
 cleanup:
230 231 232 233 234 235 236
    virBufferFreeAndReset(&buf);
    VIR_FREE(xml);
    VIR_FREE(net_xml);
    VIR_FREE(dom_xml);
    return ret;
}

237
static char *
238 239
networkDnsmasqLeaseFileNameDefault(virNetworkDriverStatePtr driver,
                                   const char *netname)
240 241 242
{
    char *leasefile;

243
    ignore_value(virAsprintf(&leasefile, "%s/%s.leases",
244
                             driver->dnsmasqStateDir, netname));
245 246 247
    return leasefile;
}

248
static char *
249 250
networkDnsmasqLeaseFileNameCustom(virNetworkDriverStatePtr driver,
                                  const char *bridge)
251 252 253 254
{
    char *leasefile;

    ignore_value(virAsprintf(&leasefile, "%s/%s.status",
255
                             driver->dnsmasqStateDir, bridge));
256 257 258
    return leasefile;
}

259
static char *
260 261
networkDnsmasqConfigFileName(virNetworkDriverStatePtr driver,
                             const char *netname)
262 263 264
{
    char *conffile;

265
    ignore_value(virAsprintf(&conffile, "%s/%s.conf",
266
                             driver->dnsmasqStateDir, netname));
267 268 269
    return conffile;
}

270 271 272 273 274 275
static char *
networkRadvdPidfileBasename(const char *netname)
{
    /* this is simple but we want to be sure it's consistently done */
    char *pidfilebase;

276
    ignore_value(virAsprintf(&pidfilebase, "%s-radvd", netname));
277 278 279 280
    return pidfilebase;
}

static char *
281 282
networkRadvdConfigFileName(virNetworkDriverStatePtr driver,
                           const char *netname)
283 284 285
{
    char *configfile;

286
    ignore_value(virAsprintf(&configfile, "%s/%s-radvd.conf",
287
                             driver->radvdStateDir, netname));
288 289
    return configfile;
}
290

291 292
/* do needed cleanup steps and remove the network from the list */
static int
293 294
networkRemoveInactive(virNetworkDriverStatePtr driver,
                      virNetworkObjPtr net)
295 296
{
    char *leasefile = NULL;
297
    char *customleasefile = NULL;
298
    char *radvdconfigfile = NULL;
299
    char *configfile = NULL;
300
    char *radvdpidbase = NULL;
301
    char *statusfile = NULL;
302 303 304 305 306 307
    dnsmasqContext *dctx = NULL;
    virNetworkDefPtr def = virNetworkObjGetPersistentDef(net);

    int ret = -1;

    /* remove the (possibly) existing dnsmasq and radvd files */
308
    if (!(dctx = dnsmasqContextNew(def->name,
309
                                   driver->dnsmasqStateDir))) {
310
        goto cleanup;
311
    }
312

313
    if (!(leasefile = networkDnsmasqLeaseFileNameDefault(driver, def->name)))
314 315
        goto cleanup;

316
    if (!(customleasefile = networkDnsmasqLeaseFileNameCustom(driver, def->bridge)))
317 318
        goto cleanup;

319
    if (!(radvdconfigfile = networkRadvdConfigFileName(driver, def->name)))
320
        goto cleanup;
321 322

    if (!(radvdpidbase = networkRadvdPidfileBasename(def->name)))
323
        goto cleanup;
324

325
    if (!(configfile = networkDnsmasqConfigFileName(driver, def->name)))
326
        goto cleanup;
327

328
    if (!(statusfile = virNetworkConfigFile(driver->stateDir, def->name)))
329
        goto cleanup;
330

331 332 333
    /* dnsmasq */
    dnsmasqDelete(dctx);
    unlink(leasefile);
334
    unlink(customleasefile);
335
    unlink(configfile);
336 337 338

    /* radvd */
    unlink(radvdconfigfile);
339
    virPidFileDelete(driver->pidDir, radvdpidbase);
340

341 342 343
    /* remove status file */
    unlink(statusfile);

344
    /* remove the network definition */
345
    virNetworkRemoveInactive(driver->networks, net);
346 347 348

    ret = 0;

349
 cleanup:
350
    VIR_FREE(leasefile);
351
    VIR_FREE(configfile);
352
    VIR_FREE(customleasefile);
353 354
    VIR_FREE(radvdconfigfile);
    VIR_FREE(radvdpidbase);
355
    VIR_FREE(statusfile);
356 357 358 359
    dnsmasqContextFree(dctx);
    return ret;
}

360 361 362
static char *
networkBridgeDummyNicName(const char *brname)
{
363
    static const char dummyNicSuffix[] = "-nic";
364 365
    char *nicname;

366 367 368 369 370 371 372
    if (strlen(brname) + sizeof(dummyNicSuffix) > IFNAMSIZ) {
        /* because the length of an ifname is limited to IFNAMSIZ-1
         * (usually 15), and we're adding 4 more characters, we must
         * truncate the original name to 11 to fit. In order to catch
         * a possible numeric ending (eg virbr0, virbr1, etc), we grab
         * the first 8 and last 3 characters of the string.
         */
373 374 375 376 377
        ignore_value(virAsprintf(&nicname, "%.*s%s%s",
                                 /* space for last 3 chars + "-nic" + NULL */
                                 (int)(IFNAMSIZ - (3 + sizeof(dummyNicSuffix))),
                                 brname, brname + strlen(brname) - 3,
                                 dummyNicSuffix));
378
    } else {
379
        ignore_value(virAsprintf(&nicname, "%s%s", brname, dummyNicSuffix));
380
    }
381 382 383
    return nicname;
}

384 385
static int
networkUpdateState(virNetworkObjPtr obj,
386
                   void *opaque)
387
{
388
    virNetworkDriverStatePtr driver = opaque;
389
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
390
    int ret = -1;
391

392
    virObjectLock(obj);
393
    if (!virNetworkObjIsActive(obj)) {
394 395
        ret = 0;
        goto cleanup;
396
    }
397

398 399 400 401 402 403 404 405
    switch (obj->def->forward.type) {
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
        /* If bridge doesn't exist, then mark it inactive */
        if (!(obj->def->bridge && virNetDevExists(obj->def->bridge) == 1))
            obj->active = 0;
        break;
406

407
    case VIR_NETWORK_FORWARD_BRIDGE:
408 409
        if (obj->def->bridge) {
            if (virNetDevExists(obj->def->bridge) != 1)
410 411
                obj->active = 0;
            break;
412
        }
413 414 415 416 417 418 419 420 421
        /* intentionally drop through to common case for all
         * macvtap networks (forward='bridge' with no bridge
         * device defined is macvtap using its 'bridge' mode)
         */
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
        /* so far no extra checks */
        break;
422

423 424 425
    case VIR_NETWORK_FORWARD_HOSTDEV:
        /* so far no extra checks */
        break;
426
    }
427

428 429 430
    /* Try and read dnsmasq/radvd pids of active networks */
    if (obj->active && obj->def->ips && (obj->def->nips > 0)) {
        char *radvdpidbase;
431

432 433 434
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
                                           obj->def->name,
                                           &obj->dnsmasqPid,
435
                                           dnsmasqCapsGetBinaryPath(dnsmasq_caps)));
436 437 438
        radvdpidbase = networkRadvdPidfileBasename(obj->def->name);
        if (!radvdpidbase)
            goto cleanup;
439

440 441 442 443
        ignore_value(virPidFileReadIfAlive(driver->pidDir,
                                           radvdpidbase,
                                           &obj->radvdPid, RADVD));
        VIR_FREE(radvdpidbase);
444
    }
445

446 447
    ret = 0;
 cleanup:
448
    virObjectUnlock(obj);
449
    virObjectUnref(dnsmasq_caps);
450 451
    return ret;
}
452

453

454 455
static int
networkAutostartConfig(virNetworkObjPtr net,
456
                       void *opaque)
457
{
458
    virNetworkDriverStatePtr driver = opaque;
459
    int ret = -1;
460

461
    virObjectLock(net);
462 463
    if (net->autostart &&
        !virNetworkObjIsActive(net) &&
464
        networkStartNetwork(driver, net) < 0)
465 466 467 468
        goto cleanup;

    ret = 0;
 cleanup:
469
    virObjectUnlock(net);
470
    return ret;
471 472
}

473 474 475
#if HAVE_FIREWALLD
static DBusHandlerResult
firewalld_dbus_filter_bridge(DBusConnection *connection ATTRIBUTE_UNUSED,
476
                             DBusMessage *message, void *user_data)
477
{
478 479
    virNetworkDriverStatePtr driver = user_data;

480 481 482 483 484 485
    if (dbus_message_is_signal(message, DBUS_INTERFACE_DBUS,
                               "NameOwnerChanged") ||
        dbus_message_is_signal(message, "org.fedoraproject.FirewallD1",
                               "Reloaded"))
    {
        VIR_DEBUG("Reload in bridge_driver because of firewalld.");
486
        networkReloadFirewallRules(driver);
487 488 489 490 491 492
    }

    return DBUS_HANDLER_RESULT_NOT_YET_HANDLED;
}
#endif

493
static int
494
networkMigrateStateFiles(virNetworkDriverStatePtr driver)
495 496 497 498 499 500 501 502 503 504 505 506 507
{
    /* Due to a change in location of network state xml beginning in
     * libvirt 1.2.4 (from /var/lib/libvirt/network to
     * /var/run/libvirt/network), we must check for state files in two
     * locations. Anything found in the old location must be written
     * to the new location, then erased from the old location. (Note
     * that we read/write the file rather than calling rename()
     * because the old and new state directories are likely in
     * different filesystems).
     */
    int ret = -1;
    const char *oldStateDir = LOCALSTATEDIR "/lib/libvirt/network";
    DIR *dir;
508
    int direrr;
509 510 511
    struct dirent *entry;
    char *oldPath = NULL, *newPath = NULL;
    char *contents = NULL;
J
Ján Tomko 已提交
512
    int rc;
513

J
Ján Tomko 已提交
514 515
    if ((rc = virDirOpenIfExists(&dir, oldStateDir)) <= 0)
        return rc;
516 517 518 519 520 521 522

    if (virFileMakePath(driver->stateDir) < 0) {
        virReportSystemError(errno, _("cannot create directory %s"),
                             driver->stateDir);
        goto cleanup;
    }

523
    while ((direrr = virDirRead(dir, &entry, oldStateDir)) > 0) {
524 525 526
        if (entry->d_type != DT_UNKNOWN &&
            entry->d_type != DT_REG)
            continue;
527 528 529

        if (virAsprintf(&oldPath, "%s/%s",
                        oldStateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
530
            goto cleanup;
531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547

        if (entry->d_type == DT_UNKNOWN) {
            struct stat st;

            if (lstat(oldPath, &st) < 0) {
                virReportSystemError(errno,
                                     _("failed to stat network status file '%s'"),
                                     oldPath);
                goto cleanup;
            }

            if (!S_ISREG(st.st_mode)) {
                VIR_FREE(oldPath);
                continue;
            }
        }

548
        if (virFileReadAll(oldPath, 1024*1024, &contents) < 0)
J
Ján Tomko 已提交
549
            goto cleanup;
550 551 552

        if (virAsprintf(&newPath, "%s/%s",
                        driver->stateDir, entry->d_name) < 0)
J
Ján Tomko 已提交
553
            goto cleanup;
554 555 556 557 558 559 560 561 562 563 564 565
        if (virFileWriteStr(newPath, contents, S_IRUSR | S_IWUSR) < 0) {
            virReportSystemError(errno,
                                 _("failed to write network status file '%s'"),
                                 newPath);
            goto cleanup;
        }

        unlink(oldPath);
        VIR_FREE(oldPath);
        VIR_FREE(newPath);
        VIR_FREE(contents);
    }
566
    if (direrr < 0)
J
Ján Tomko 已提交
567
        goto cleanup;
568 569 570

    ret = 0;
 cleanup:
J
Ján Tomko 已提交
571
    VIR_DIR_CLOSE(dir);
572 573 574 575 576 577
    VIR_FREE(oldPath);
    VIR_FREE(newPath);
    VIR_FREE(contents);
    return ret;
}

578
/**
579
 * networkStateInitialize:
580 581 582 583
 *
 * Initialization function for the QEmu daemon
 */
static int
584 585 586
networkStateInitialize(bool privileged,
                       virStateInhibitCallback callback ATTRIBUTE_UNUSED,
                       void *opaque ATTRIBUTE_UNUSED)
587
{
588 589 590
    int ret = -1;
    char *configdir = NULL;
    char *rundir = NULL;
591 592 593
#ifdef HAVE_FIREWALLD
    DBusConnection *sysbus = NULL;
#endif
594

595
    if (VIR_ALLOC(network_driver) < 0)
596
        goto error;
597

598 599
    if (virMutexInit(&network_driver->lock) < 0) {
        VIR_FREE(network_driver);
600 601
        goto error;
    }
602

603 604 605 606
    /* configuration/state paths are one of
     * ~/.config/libvirt/... (session/unprivileged)
     * /etc/libvirt/... && /var/(run|lib)/libvirt/... (system/privileged).
     */
607
    if (privileged) {
608
        if (VIR_STRDUP(network_driver->networkConfigDir,
609
                       SYSCONFDIR "/libvirt/qemu/networks") < 0 ||
610
            VIR_STRDUP(network_driver->networkAutostartDir,
611
                       SYSCONFDIR "/libvirt/qemu/networks/autostart") < 0 ||
612
            VIR_STRDUP(network_driver->stateDir,
613
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
614
            VIR_STRDUP(network_driver->pidDir,
615
                       LOCALSTATEDIR "/run/libvirt/network") < 0 ||
616
            VIR_STRDUP(network_driver->dnsmasqStateDir,
617
                       LOCALSTATEDIR "/lib/libvirt/dnsmasq") < 0 ||
618
            VIR_STRDUP(network_driver->radvdStateDir,
619 620
                       LOCALSTATEDIR "/lib/libvirt/radvd") < 0)
            goto error;
621 622 623 624 625

        /* migration from old to new location is only applicable for
         * privileged mode - unprivileged mode directories haven't
         * changed location.
         */
626
        if (networkMigrateStateFiles(network_driver) < 0)
627
            goto error;
628
    } else {
629 630 631
        configdir = virGetUserConfigDirectory();
        rundir = virGetUserRuntimeDirectory();
        if (!(configdir && rundir))
632
            goto error;
633

634
        if ((virAsprintf(&network_driver->networkConfigDir,
635
                         "%s/qemu/networks", configdir) < 0) ||
636
            (virAsprintf(&network_driver->networkAutostartDir,
637
                         "%s/qemu/networks/autostart", configdir) < 0) ||
638
            (virAsprintf(&network_driver->stateDir,
639
                         "%s/network/lib", rundir) < 0) ||
640
            (virAsprintf(&network_driver->pidDir,
641
                         "%s/network/run", rundir) < 0) ||
642
            (virAsprintf(&network_driver->dnsmasqStateDir,
643
                         "%s/dnsmasq/lib", rundir) < 0) ||
644
            (virAsprintf(&network_driver->radvdStateDir,
645
                         "%s/radvd/lib", rundir) < 0)) {
646
            goto error;
647
        }
648 649
    }

650
    if (virFileMakePath(network_driver->stateDir) < 0) {
651 652
        virReportSystemError(errno,
                             _("cannot create directory %s"),
653
                             network_driver->stateDir);
654 655 656
        goto error;
    }

657
    /* if this fails now, it will be retried later with dnsmasqCapsRefresh() */
658
    network_driver->dnsmasqCaps = dnsmasqCapsNewFromBinary(DNSMASQ);
659

660
    if (!(network_driver->networks = virNetworkObjListNew()))
661 662
        goto error;

663 664
    if (virNetworkLoadAllState(network_driver->networks,
                               network_driver->stateDir) < 0)
665 666
        goto error;

667 668 669
    if (virNetworkLoadAllConfigs(network_driver->networks,
                                 network_driver->networkConfigDir,
                                 network_driver->networkAutostartDir) < 0)
670 671
        goto error;

672 673 674 675
    /* Update the internal status of all allegedly active
     * networks according to external conditions on the host
     * (i.e. anything that isn't stored directly in each
     * network's state file). */
676
    virNetworkObjListForEach(network_driver->networks,
677
                             networkUpdateState,
678 679
                             network_driver);
    virNetworkObjListPrune(network_driver->networks,
680 681
                           VIR_CONNECT_LIST_NETWORKS_INACTIVE |
                           VIR_CONNECT_LIST_NETWORKS_TRANSIENT);
682 683
    networkReloadFirewallRules(network_driver);
    networkRefreshDaemons(network_driver);
684

685
    network_driver->networkEventState = virObjectEventStateNew();
686

687 688 689
#ifdef HAVE_FIREWALLD
    if (!(sysbus = virDBusGetSystemBus())) {
        VIR_WARN("DBus not available, disabling firewalld support "
690
                 "in bridge_network_driver: %s", virGetLastErrorMessage());
691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707
    } else {
        /* add matches for
         * NameOwnerChanged on org.freedesktop.DBus for firewalld start/stop
         * Reloaded on org.fedoraproject.FirewallD1 for firewalld reload
         */
        dbus_bus_add_match(sysbus,
                           "type='signal'"
                           ",interface='"DBUS_INTERFACE_DBUS"'"
                           ",member='NameOwnerChanged'"
                           ",arg0='org.fedoraproject.FirewallD1'",
                           NULL);
        dbus_bus_add_match(sysbus,
                           "type='signal'"
                           ",interface='org.fedoraproject.FirewallD1'"
                           ",member='Reloaded'",
                           NULL);
        dbus_connection_add_filter(sysbus, firewalld_dbus_filter_bridge,
708
                                   network_driver, NULL);
709 710 711
    }
#endif

712
    ret = 0;
713
 cleanup:
714 715 716
    VIR_FREE(configdir);
    VIR_FREE(rundir);
    return ret;
717

718
 error:
719 720
    if (network_driver)
        networkDriverUnlock(network_driver);
721
    networkStateCleanup();
722
    goto cleanup;
723 724
}

725 726 727 728 729 730 731 732
/**
 * networkStateAutoStart:
 *
 * Function to AutoStart the bridge configs
 */
static void
networkStateAutoStart(void)
{
733
    if (!network_driver)
734 735
        return;

736
    virNetworkObjListForEach(network_driver->networks,
737
                             networkAutostartConfig,
738
                             network_driver);
739 740
}

741
/**
742
 * networkStateReload:
743 744 745 746 747
 *
 * Function to restart the QEmu daemon, it will recheck the configuration
 * files and update its state and the networking
 */
static int
748 749
networkStateReload(void)
{
750
    if (!network_driver)
751 752
        return 0;

753 754 755 756 757 758 759 760
    virNetworkLoadAllState(network_driver->networks,
                           network_driver->stateDir);
    virNetworkLoadAllConfigs(network_driver->networks,
                             network_driver->networkConfigDir,
                             network_driver->networkAutostartDir);
    networkReloadFirewallRules(network_driver);
    networkRefreshDaemons(network_driver);
    virNetworkObjListForEach(network_driver->networks,
761
                             networkAutostartConfig,
762
                             network_driver);
763 764 765 766 767
    return 0;
}


/**
768
 * networkStateCleanup:
769 770 771 772
 *
 * Shutdown the QEmu daemon, it will stop all active domains and networks
 */
static int
773 774
networkStateCleanup(void)
{
775
    if (!network_driver)
776 777
        return -1;

778
    virObjectEventStateFree(network_driver->networkEventState);
779

780
    /* free inactive networks */
781
    virObjectUnref(network_driver->networks);
782

783 784 785 786 787 788
    VIR_FREE(network_driver->networkConfigDir);
    VIR_FREE(network_driver->networkAutostartDir);
    VIR_FREE(network_driver->stateDir);
    VIR_FREE(network_driver->pidDir);
    VIR_FREE(network_driver->dnsmasqStateDir);
    VIR_FREE(network_driver->radvdStateDir);
789

790
    virObjectUnref(network_driver->dnsmasqCaps);
791

792
    virMutexDestroy(&network_driver->lock);
793

794
    VIR_FREE(network_driver);
795 796 797 798 799

    return 0;
}


800 801 802 803 804 805 806
/* networkKillDaemon:
 *
 * kill the specified pid/name, and wait a bit to make sure it's dead.
 */
static int
networkKillDaemon(pid_t pid, const char *daemonName, const char *networkName)
{
807 808
    size_t i;
    int ret = -1;
809 810 811 812 813 814 815
    const char *signame = "TERM";

    /* send SIGTERM, then wait up to 3 seconds for the process to
     * disappear, send SIGKILL, then wait for up to another 2
     * seconds. If that fails, log a warning and continue, hoping
     * for the best.
     */
816
    for (i = 0; i < 25; i++) {
817
        int signum = 0;
818
        if (i == 0) {
819
            signum = SIGTERM;
820
        } else if (i == 15) {
821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855
            signum = SIGKILL;
            signame = "KILL";
        }
        if (kill(pid, signum) < 0) {
            if (errno == ESRCH) {
                ret = 0;
            } else {
                char ebuf[1024];
                VIR_WARN("Failed to terminate %s process %d "
                         "for network '%s' with SIG%s: %s",
                         daemonName, pid, networkName, signame,
                         virStrerror(errno, ebuf, sizeof(ebuf)));
            }
            goto cleanup;
        }
        /* NB: since networks have no reference count like
         * domains, there is no safe way to unlock the network
         * object temporarily, and so we can't follow the
         * procedure used by the qemu driver of 1) unlock driver
         * 2) sleep, 3) add ref to object 4) unlock object, 5)
         * re-lock driver, 6) re-lock object. We may need to add
         * that functionality eventually, but for now this
         * function is rarely used and, at worst, leaving the
         * network driver locked during this loop of sleeps will
         * have the effect of holding up any other thread trying
         * to make modifications to a network for up to 5 seconds;
         * since modifications to networks are much less common
         * than modifications to domains, this seems a reasonable
         * tradeoff in exchange for less code disruption.
         */
        usleep(20 * 1000);
    }
    VIR_WARN("Timed out waiting after SIG%s to %s process %d "
             "(network '%s')",
             signame, daemonName, pid, networkName);
856
 cleanup:
857 858 859
    return ret;
}

J
Ján Tomko 已提交
860 861 862
/* the following does not build a file, it builds a list
 * which is later saved into a file
 */
G
Gene Czarcinski 已提交
863

864
static int
G
Gene Czarcinski 已提交
865
networkBuildDnsmasqDhcpHostsList(dnsmasqContext *dctx,
866
                                 virNetworkIPDefPtr ipdef)
867
{
868
    size_t i;
G
Gene Czarcinski 已提交
869
    bool ipv6 = false;
870

G
Gene Czarcinski 已提交
871 872
    if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
        ipv6 = true;
873 874
    for (i = 0; i < ipdef->nhosts; i++) {
        virNetworkDHCPHostDefPtr host = &(ipdef->hosts[i]);
G
Gene Czarcinski 已提交
875
        if (VIR_SOCKET_ADDR_VALID(&host->ip))
876 877
            if (dnsmasqAddDhcpHost(dctx, host->mac, &host->ip,
                                   host->name, host->id, ipv6) < 0)
878
                return -1;
879
    }
880

G
Gene Czarcinski 已提交
881 882 883 884 885 886 887
    return 0;
}

static int
networkBuildDnsmasqHostsList(dnsmasqContext *dctx,
                             virNetworkDNSDefPtr dnsdef)
{
888
    size_t i, j;
G
Gene Czarcinski 已提交
889

890 891
    if (dnsdef) {
        for (i = 0; i < dnsdef->nhosts; i++) {
892
            virNetworkDNSHostDefPtr host = &(dnsdef->hosts[i]);
893
            if (VIR_SOCKET_ADDR_VALID(&host->ip)) {
894
                for (j = 0; j < host->nnames; j++)
895 896
                    if (dnsmasqAddHost(dctx, &host->ip, host->names[j]) < 0)
                        return -1;
897 898
            }
        }
899 900
    }

901
    return 0;
902 903 904
}


905 906
int
networkDnsmasqConfContents(virNetworkObjPtr network,
907 908 909 910
                           const char *pidfile,
                           char **configstr,
                           dnsmasqContext *dctx,
                           dnsmasqCapsPtr caps ATTRIBUTE_UNUSED)
911
{
912
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
913
    int r, ret = -1;
914
    int nbleases = 0;
915
    size_t i;
916
    virNetworkDNSDefPtr dns = &network->def->dns;
917
    virNetworkIPDefPtr tmpipdef, ipdef, ipv4def, ipv6def;
G
Gene Czarcinski 已提交
918
    bool ipv6SLAAC;
919
    char *saddr = NULL, *eaddr = NULL;
920

921 922
    *configstr = NULL;

923
    /*
924 925 926
     * All dnsmasq parameters are put into a configuration file, except the
     * command line --conf-file=parameter which specifies the location of
     * configuration file.
927
     *
928 929
     * All dnsmasq conf-file parameters must be specified as "foo=bar"
     * as oppose to "--foo bar" which was acceptable on the command line.
930
     */
931 932 933 934 935 936

    /*
     * Needed to ensure dnsmasq uses same algorithm for processing
     * multiple namedriver entries in /etc/resolv.conf as GLibC.
     */

937 938
    /* create dnsmasq config file appropriate for this network */
    virBufferAsprintf(&configbuf,
939 940 941 942 943 944 945
                      "##WARNING:  THIS IS AN AUTO-GENERATED FILE. "
                      "CHANGES TO IT ARE LIKELY TO BE\n"
                      "##OVERWRITTEN AND LOST.  Changes to this "
                      "configuration should be made using:\n"
                      "##    virsh net-edit %s\n"
                      "## or other application using the libvirt API.\n"
                      "##\n## dnsmasq conf file created by libvirt\n"
946
                      "strict-order\n",
947 948
                      network->def->name);

949 950 951 952
    if (network->def->dns.forwarders) {
        virBufferAddLit(&configbuf, "no-resolv\n");
        for (i = 0; i < network->def->dns.nfwds; i++) {
            virBufferAsprintf(&configbuf, "server=%s\n",
J
Ján Tomko 已提交
953
                              network->def->dns.forwarders[i]);
954 955 956
        }
    }

957
    if (network->def->domain) {
958 959 960 961 962
        if (network->def->domainLocalOnly == VIR_TRISTATE_BOOL_YES) {
            virBufferAsprintf(&configbuf,
                              "local=/%s/\n",
                              network->def->domain);
        }
963
        virBufferAsprintf(&configbuf,
964 965 966 967
                          "domain=%s\n"
                          "expand-hosts\n",
                          network->def->domain);
    }
968

J
Ján Tomko 已提交
969
    if (network->def->dns.forwardPlainNames == VIR_TRISTATE_BOOL_NO) {
970 971 972 973
        virBufferAddLit(&configbuf, "domain-needed\n");
        /* need to specify local=// whether or not a domain is
         * specified, unless the config says we should forward "plain"
         * names (i.e. not fully qualified, no '.' characters)
974
         */
975
        virBufferAddLit(&configbuf, "local=//\n");
976
    }
977

978
    if (pidfile)
979
        virBufferAsprintf(&configbuf, "pid-file=%s\n", pidfile);
980

981 982 983
    /* dnsmasq will *always* listen on localhost unless told otherwise */
    virBufferAddLit(&configbuf, "except-interface=lo\n");

984 985 986 987 988 989 990 991
    if (dnsmasqCapsGet(caps, DNSMASQ_CAPS_BIND_DYNAMIC)) {
        /* using --bind-dynamic with only --interface (no
         * --listen-address) prevents dnsmasq from responding to dns
         * queries that arrive on some interface other than our bridge
         * interface (in other words, requests originating somewhere
         * other than one of the virtual guests connected directly to
         * this network). This was added in response to CVE 2012-3411.
         */
992
        virBufferAsprintf(&configbuf,
993 994 995
                          "bind-dynamic\n"
                          "interface=%s\n",
                          network->def->bridge);
996
    } else {
997
        virBufferAddLit(&configbuf, "bind-interfaces\n");
998 999 1000 1001 1002 1003 1004 1005
        /*
         * --interface does not actually work with dnsmasq < 2.47,
         * due to DAD for ipv6 addresses on the interface.
         *
         * virCommandAddArgList(cmd, "--interface", network->def->bridge, NULL);
         *
         * So listen on all defined IPv[46] addresses
         */
1006
        for (i = 0;
1007
             (tmpipdef = virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, i));
1008
             i++) {
1009 1010 1011 1012
            char *ipaddr = virSocketAddrFormat(&tmpipdef->address);

            if (!ipaddr)
                goto cleanup;
1013

1014
            /* also part of CVE 2012-3411 - if the host's version of
1015
             * dnsmasq doesn't have bind-dynamic, only allow listening on
1016 1017
             * private/local IP addresses (see RFC1918/RFC3484/RFC4193)
             */
1018 1019
            if (!dnsmasqCapsGet(caps, DNSMASQ_CAPS_BINDTODEVICE) &&
                !virSocketAddrIsPrivate(&tmpipdef->address)) {
1020 1021 1022 1023
                unsigned long version = dnsmasqCapsGetVersion(caps);

                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("Publicly routable address %s is prohibited. "
1024
                                 "The version of dnsmasq on this host (%d.%d) "
1025 1026 1027 1028
                                 "doesn't support the bind-dynamic option or "
                                 "use SO_BINDTODEVICE on listening sockets, "
                                 "one of which is required for safe operation "
                                 "on a publicly routable subnet "
1029 1030 1031 1032 1033 1034
                                 "(see CVE-2012-3411). You must either "
                                 "upgrade dnsmasq, or use a private/local "
                                 "subnet range for this network "
                                 "(as described in RFC1918/RFC3484/RFC4193)."),
                               ipaddr, (int)version / 1000000,
                               (int)(version % 1000000) / 1000);
1035
                VIR_FREE(ipaddr);
1036 1037
                goto cleanup;
            }
1038
            virBufferAsprintf(&configbuf, "listen-address=%s\n", ipaddr);
1039 1040 1041
            VIR_FREE(ipaddr);
        }
    }
1042

1043 1044
    /* If this is an isolated network, set the default route option
     * (3) to be empty to avoid setting a default route that's
1045
     * guaranteed to not work, and set no-resolv so that no dns
1046 1047 1048
     * requests are forwarded on to the dns server listed in the
     * host's /etc/resolv.conf (since this could be used as a channel
     * to build a connection to the outside).
1049
     */
1050
    if (network->def->forward.type == VIR_NETWORK_FORWARD_NONE) {
1051
        virBufferAddLit(&configbuf, "dhcp-option=3\n"
1052
                        "no-resolv\n");
1053
    }
1054

1055
    for (i = 0; i < dns->ntxts; i++) {
1056
        virBufferAsprintf(&configbuf, "txt-record=%s,%s\n",
1057 1058
                          dns->txts[i].name,
                          dns->txts[i].value);
1059
    }
1060

1061
    for (i = 0; i < dns->nsrvs; i++) {
1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083
        /* service/protocol are required, and should have been validated
         * by the parser.
         */
        if (!dns->srvs[i].service) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Missing required 'service' "
                             "attribute in SRV record of network '%s'"),
                           network->def->name);
            goto cleanup;
        }
        if (!dns->srvs[i].protocol) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("Missing required 'service' "
                             "attribute in SRV record of network '%s'"),
                           network->def->name);
            goto cleanup;
        }
        /* RFC2782 requires that service and protocol be preceded by
         * an underscore.
         */
        virBufferAsprintf(&configbuf, "srv-host=_%s._%s",
                          dns->srvs[i].service, dns->srvs[i].protocol);
1084

1085 1086 1087
        /* domain is optional - it defaults to the domain of this network */
        if (dns->srvs[i].domain)
            virBufferAsprintf(&configbuf, ".%s", dns->srvs[i].domain);
1088

1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110
        /* If target is empty or ".", that means "the service is
         * decidedly not available at this domain" (RFC2782). In that
         * case, any port, priority, or weight is irrelevant.
         */
        if (dns->srvs[i].target && STRNEQ(dns->srvs[i].target, ".")) {

            virBufferAsprintf(&configbuf, ",%s", dns->srvs[i].target);
            /* port, priority, and weight are optional, but are
             * identified by their position in the line. If an item is
             * unspecified, but something later in the line *is*
             * specified, we need to give the default value for the
             * unspecified item. (According to the dnsmasq manpage,
             * the default for port is 1).
             */
            if (dns->srvs[i].port ||
                dns->srvs[i].priority || dns->srvs[i].weight)
                virBufferAsprintf(&configbuf, ",%d",
                                  dns->srvs[i].port ? dns->srvs[i].port : 1);
            if (dns->srvs[i].priority || dns->srvs[i].weight)
                virBufferAsprintf(&configbuf, ",%d", dns->srvs[i].priority);
            if (dns->srvs[i].weight)
                virBufferAsprintf(&configbuf, ",%d", dns->srvs[i].weight);
1111
        }
1112
        virBufferAddLit(&configbuf, "\n");
1113 1114
    }

G
Gene Czarcinski 已提交
1115
    /* Find the first dhcp for both IPv4 and IPv6 */
1116
    for (i = 0, ipv4def = NULL, ipv6def = NULL, ipv6SLAAC = false;
1117
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, i));
1118
         i++) {
G
Gene Czarcinski 已提交
1119 1120 1121 1122
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1123 1124
                                   _("For IPv4, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135
                    goto cleanup;
                } else {
                    ipv4def = ipdef;
                }
            }
        }
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (!DNSMASQ_DHCPv6_SUPPORT(caps)) {
                    unsigned long version = dnsmasqCapsGetVersion(caps);
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
1136 1137 1138 1139 1140 1141 1142 1143 1144
                                   _("The version of dnsmasq on this host "
                                     "(%d.%d) doesn't adequately support "
                                     "IPv6 dhcp range or dhcp host "
                                     "specification. Version %d.%d or later "
                                     "is required."),
                                   (int)version / 1000000,
                                   (int)(version % 1000000) / 1000,
                                   DNSMASQ_DHCPv6_MAJOR_REQD,
                                   DNSMASQ_DHCPv6_MINOR_REQD);
G
Gene Czarcinski 已提交
1145 1146 1147 1148
                    goto cleanup;
                }
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
1149 1150
                                   _("For IPv6, multiple DHCP definitions "
                                     "cannot be specified."));
G
Gene Czarcinski 已提交
1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163
                    goto cleanup;
                } else {
                    ipv6def = ipdef;
                }
            } else {
                ipv6SLAAC = true;
            }
        }
    }

    if (ipv6def && ipv6SLAAC) {
        VIR_WARN("For IPv6, when DHCP is specified for one address, then "
                 "state-full Router Advertising will occur.  The additional "
1164 1165 1166 1167
                 "IPv6 addresses specified require manually configured guest "
                 "network to work properly since both state-full (DHCP) "
                 "and state-less (SLAAC) addressing are not supported "
                 "on the same network interface.");
G
Gene Czarcinski 已提交
1168 1169 1170 1171 1172
    }

    ipdef = ipv4def ? ipv4def : ipv6def;

    while (ipdef) {
1173 1174
        int prefix;

1175
        prefix = virNetworkIPDefPrefix(ipdef);
1176 1177 1178 1179 1180 1181
        if (prefix < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
            goto cleanup;
        }
1182
        for (r = 0; r < ipdef->nranges; r++) {
1183 1184
            int thisRange;

1185 1186
            if (!(saddr = virSocketAddrFormat(&ipdef->ranges[r].start)) ||
                !(eaddr = virSocketAddrFormat(&ipdef->ranges[r].end)))
1187
                goto cleanup;
1188

1189
            virBufferAsprintf(&configbuf, "dhcp-range=%s,%s",
1190
                              saddr, eaddr);
1191 1192 1193 1194
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
               virBufferAsprintf(&configbuf, ",%d", prefix);
            virBufferAddLit(&configbuf, "\n");

1195
            VIR_FREE(saddr);
1196
            VIR_FREE(eaddr);
1197
            thisRange = virSocketAddrGetRange(&ipdef->ranges[r].start,
1198 1199
                                              &ipdef->ranges[r].end,
                                              &ipdef->address,
1200
                                              virNetworkIPDefPrefix(ipdef));
1201 1202 1203
            if (thisRange < 0)
                goto cleanup;
            nbleases += thisRange;
1204
        }
1205

1206
        /*
1207 1208 1209 1210
         * For static-only DHCP, i.e. with no range but at least one
         * host element, we have to add a special --dhcp-range option
         * to enable the service in dnsmasq. (this is for dhcp-hosts=
         * support)
1211 1212
         */
        if (!ipdef->nranges && ipdef->nhosts) {
1213
            char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
1214 1215
            if (!bridgeaddr)
                goto cleanup;
1216 1217 1218 1219 1220
            virBufferAsprintf(&configbuf, "dhcp-range=%s,static",
                              bridgeaddr);
            if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
               virBufferAsprintf(&configbuf, ",%d", prefix);
            virBufferAddLit(&configbuf, "\n");
1221 1222
            VIR_FREE(bridgeaddr);
        }
1223

G
Gene Czarcinski 已提交
1224 1225
        if (networkBuildDnsmasqDhcpHostsList(dctx, ipdef) < 0)
            goto cleanup;
1226

G
Gene Czarcinski 已提交
1227 1228 1229
        /* Note: the following is IPv4 only */
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts)
1230
                virBufferAddLit(&configbuf, "dhcp-no-override\n");
1231

G
Gene Czarcinski 已提交
1232
            if (ipdef->tftproot) {
1233 1234
                virBufferAddLit(&configbuf, "enable-tftp\n");
                virBufferAsprintf(&configbuf, "tftp-root=%s\n", ipdef->tftproot);
G
Gene Czarcinski 已提交
1235
            }
1236

G
Gene Czarcinski 已提交
1237 1238 1239
            if (ipdef->bootfile) {
                if (VIR_SOCKET_ADDR_VALID(&ipdef->bootserver)) {
                    char *bootserver = virSocketAddrFormat(&ipdef->bootserver);
1240

1241
                    if (!bootserver)
G
Gene Czarcinski 已提交
1242
                        goto cleanup;
1243
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s%s%s\n",
1244
                                      ipdef->bootfile, ",,", bootserver);
G
Gene Czarcinski 已提交
1245 1246
                    VIR_FREE(bootserver);
                } else {
1247
                    virBufferAsprintf(&configbuf, "dhcp-boot=%s\n", ipdef->bootfile);
G
Gene Czarcinski 已提交
1248 1249 1250 1251 1252
                }
            }
        }
        ipdef = (ipdef == ipv6def) ? NULL : ipv6def;
    }
1253

1254
    if (nbleases > 0)
1255
        virBufferAsprintf(&configbuf, "dhcp-lease-max=%d\n", nbleases);
1256

G
Gene Czarcinski 已提交
1257 1258
    /* this is done once per interface */
    if (networkBuildDnsmasqHostsList(dctx, dns) < 0)
1259
        goto cleanup;
G
Gene Czarcinski 已提交
1260 1261 1262 1263 1264 1265

    /* Even if there are currently no static hosts, if we're
     * listening for DHCP, we should write a 0-length hosts
     * file to allow for runtime additions.
     */
    if (ipv4def || ipv6def)
1266 1267
        virBufferAsprintf(&configbuf, "dhcp-hostsfile=%s\n",
                          dctx->hostsfile->path);
G
Gene Czarcinski 已提交
1268

1269 1270
    /* Likewise, always create this file and put it on the
     * commandline, to allow for runtime additions.
G
Gene Czarcinski 已提交
1271
     */
1272
    virBufferAsprintf(&configbuf, "addn-hosts=%s\n",
1273
                      dctx->addnhostsfile->path);
G
Gene Czarcinski 已提交
1274 1275 1276

    /* Are we doing RA instead of radvd? */
    if (DNSMASQ_RA_SUPPORT(caps)) {
1277
        if (ipv6def) {
1278
            virBufferAddLit(&configbuf, "enable-ra\n");
1279
        } else {
1280
            for (i = 0;
1281
                 (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET6, i));
1282
                 i++) {
G
Gene Czarcinski 已提交
1283 1284 1285 1286
                if (!(ipdef->nranges || ipdef->nhosts)) {
                    char *bridgeaddr = virSocketAddrFormat(&ipdef->address);
                    if (!bridgeaddr)
                        goto cleanup;
1287 1288
                    virBufferAsprintf(&configbuf,
                                      "dhcp-range=%s,ra-only\n", bridgeaddr);
G
Gene Czarcinski 已提交
1289 1290
                    VIR_FREE(bridgeaddr);
                }
1291
            }
1292
        }
1293 1294
    }

1295 1296 1297
    if (!(*configstr = virBufferContentAndReset(&configbuf)))
        goto cleanup;

1298
    ret = 0;
G
Gene Czarcinski 已提交
1299

1300
 cleanup:
1301 1302
    VIR_FREE(saddr);
    VIR_FREE(eaddr);
1303
    virBufferFreeAndReset(&configbuf);
1304
    return ret;
1305 1306
}

1307
/* build the dnsmasq command line */
1308 1309 1310
static int ATTRIBUTE_NONNULL(3)
networkBuildDhcpDaemonCommandLine(virNetworkDriverStatePtr driver,
                                  virNetworkObjPtr network,
1311
                                  virCommandPtr *cmdout,
1312 1313
                                  char *pidfile,
                                  dnsmasqContext *dctx)
1314
{
1315
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1316
    virCommandPtr cmd = NULL;
G
Gene Czarcinski 已提交
1317
    int ret = -1;
1318 1319
    char *configfile = NULL;
    char *configstr = NULL;
1320
    char *leaseshelper_path = NULL;
1321 1322

    network->dnsmasqPid = -1;
1323

1324
    if (networkDnsmasqConfContents(network, pidfile, &configstr,
1325
                                   dctx, dnsmasq_caps) < 0)
1326 1327 1328 1329 1330
        goto cleanup;
    if (!configstr)
        goto cleanup;

    /* construct the filename */
1331
    if (!(configfile = networkDnsmasqConfigFileName(driver, network->def->name)))
1332 1333 1334 1335 1336
        goto cleanup;

    /* Write the file */
    if (virFileWriteStr(configfile, configstr, 0600) < 0) {
        virReportSystemError(errno,
J
Ján Tomko 已提交
1337 1338
                             _("couldn't write dnsmasq config file '%s'"),
                             configfile);
1339 1340 1341
        goto cleanup;
    }

1342 1343
    /* This helper is used to create custom leases file for libvirt */
    if (!(leaseshelper_path = virFileFindResource("libvirt_leaseshelper",
1344
                                                  abs_topbuilddir "/src",
1345 1346 1347
                                                  LIBEXECDIR)))
        goto cleanup;

1348
    cmd = virCommandNew(dnsmasqCapsGetBinaryPath(dnsmasq_caps));
1349
    virCommandAddArgFormat(cmd, "--conf-file=%s", configfile);
1350 1351
    /* Libvirt gains full control of leases database */
    virCommandAddArgFormat(cmd, "--leasefile-ro");
1352
    virCommandAddArgFormat(cmd, "--dhcp-script=%s", leaseshelper_path);
1353
    virCommandAddEnvPair(cmd, "VIR_BRIDGE_NAME", network->def->bridge);
1354

1355
    *cmdout = cmd;
1356
    ret = 0;
1357
 cleanup:
1358
    virObjectUnref(dnsmasq_caps);
1359 1360
    VIR_FREE(configfile);
    VIR_FREE(configstr);
1361
    VIR_FREE(leaseshelper_path);
1362 1363 1364 1365
    return ret;
}

static int
1366 1367
networkStartDhcpDaemon(virNetworkDriverStatePtr driver,
                       virNetworkObjPtr network)
1368 1369 1370 1371
{
    virCommandPtr cmd = NULL;
    char *pidfile = NULL;
    int ret = -1;
1372
    dnsmasqContext *dctx = NULL;
1373

1374
    if (!virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, 0)) {
G
Gene Czarcinski 已提交
1375
        /* no IP addresses, so we don't need to run */
1376 1377 1378 1379
        ret = 0;
        goto cleanup;
    }

1380
    if (virFileMakePath(driver->pidDir) < 0) {
1381
        virReportSystemError(errno,
1382
                             _("cannot create directory %s"),
1383
                             driver->pidDir);
1384
        goto cleanup;
1385 1386
    }

1387
    if (!(pidfile = virPidFileBuildPath(driver->pidDir,
1388
                                        network->def->name)))
1389
        goto cleanup;
1390

1391
    if (virFileMakePath(driver->dnsmasqStateDir) < 0) {
1392
        virReportSystemError(errno,
1393
                             _("cannot create directory %s"),
1394
                             driver->dnsmasqStateDir);
1395 1396 1397
        goto cleanup;
    }

1398
    dctx = dnsmasqContextNew(network->def->name, driver->dnsmasqStateDir);
1399 1400 1401
    if (dctx == NULL)
        goto cleanup;

1402
    if (networkDnsmasqCapsRefresh(driver) < 0)
1403
        goto cleanup;
1404

1405
    ret = networkBuildDhcpDaemonCommandLine(driver, network, &cmd, pidfile, dctx);
1406 1407 1408 1409 1410
    if (ret < 0)
        goto cleanup;

    ret = dnsmasqSave(dctx);
    if (ret < 0)
1411
        goto cleanup;
1412

G
Guido Günther 已提交
1413
    ret = virCommandRun(cmd, NULL);
1414
    if (ret < 0)
1415 1416 1417
        goto cleanup;

    /*
1418 1419 1420 1421 1422
     * There really is no race here - when dnsmasq daemonizes, its
     * leader process stays around until its child has actually
     * written its pidfile. So by time virCommandRun exits it has
     * waitpid'd and guaranteed the proess has started and written a
     * pid
1423 1424
     */

1425
    ret = virPidFileRead(driver->pidDir, network->def->name,
1426 1427
                         &network->dnsmasqPid);
    if (ret < 0)
1428
        goto cleanup;
1429

1430
    ret = 0;
1431
 cleanup:
1432
    VIR_FREE(pidfile);
1433
    virCommandFree(cmd);
1434
    dnsmasqContextFree(dctx);
1435 1436 1437
    return ret;
}

1438 1439
/* networkRefreshDhcpDaemon:
 *  Update dnsmasq config files, then send a SIGHUP so that it rereads
G
Gene Czarcinski 已提交
1440 1441
 *  them.   This only works for the dhcp-hostsfile and the
 *  addn-hosts file.
1442 1443 1444
 *
 *  Returns 0 on success, -1 on failure.
 */
1445
static int
1446 1447
networkRefreshDhcpDaemon(virNetworkDriverStatePtr driver,
                         virNetworkObjPtr network)
1448
{
1449 1450
    int ret = -1;
    size_t i;
1451
    virNetworkIPDefPtr ipdef, ipv4def, ipv6def;
1452
    dnsmasqContext *dctx = NULL;
1453

G
Gene Czarcinski 已提交
1454
    /* if no IP addresses specified, nothing to do */
1455
    if (!virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, 0))
G
Gene Czarcinski 已提交
1456 1457
        return 0;

1458 1459
    /* if there's no running dnsmasq, just start it */
    if (network->dnsmasqPid <= 0 || (kill(network->dnsmasqPid, 0) < 0))
1460
        return networkStartDhcpDaemon(driver, network);
1461

G
Gene Czarcinski 已提交
1462
    VIR_INFO("Refreshing dnsmasq for network %s", network->def->bridge);
1463
    if (!(dctx = dnsmasqContextNew(network->def->name,
1464
                                   driver->dnsmasqStateDir))) {
G
Gene Czarcinski 已提交
1465
        goto cleanup;
1466
    }
G
Gene Czarcinski 已提交
1467 1468 1469 1470 1471 1472

    /* Look for first IPv4 address that has dhcp defined.
     * We only support dhcp-host config on one IPv4 subnetwork
     * and on one IPv6 subnetwork.
     */
    ipv4def = NULL;
1473
    for (i = 0;
1474
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET, i));
1475
         i++) {
G
Gene Czarcinski 已提交
1476 1477
        if (!ipv4def && (ipdef->nranges || ipdef->nhosts))
            ipv4def = ipdef;
1478 1479
    }

G
Gene Czarcinski 已提交
1480
    ipv6def = NULL;
1481
    for (i = 0;
1482
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET6, i));
1483
         i++) {
G
Gene Czarcinski 已提交
1484 1485
        if (!ipv6def && (ipdef->nranges || ipdef->nhosts))
            ipv6def = ipdef;
1486 1487
    }

G
Gene Czarcinski 已提交
1488
    if (ipv4def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv4def) < 0))
J
Ján Tomko 已提交
1489
        goto cleanup;
G
Gene Czarcinski 已提交
1490 1491

    if (ipv6def && (networkBuildDnsmasqDhcpHostsList(dctx, ipv6def) < 0))
J
Ján Tomko 已提交
1492
        goto cleanup;
1493

G
Gene Czarcinski 已提交
1494
    if (networkBuildDnsmasqHostsList(dctx, &network->def->dns) < 0)
J
Ján Tomko 已提交
1495
        goto cleanup;
1496 1497

    if ((ret = dnsmasqSave(dctx)) < 0)
1498
        goto cleanup;
1499 1500

    ret = kill(network->dnsmasqPid, SIGHUP);
1501
 cleanup:
1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513
    dnsmasqContextFree(dctx);
    return ret;
}

/* networkRestartDhcpDaemon:
 *
 * kill and restart dnsmasq, in order to update any config that is on
 * the dnsmasq commandline (and any placed in separate config files).
 *
 *  Returns 0 on success, -1 on failure.
 */
static int
1514 1515
networkRestartDhcpDaemon(virNetworkDriverStatePtr driver,
                         virNetworkObjPtr network)
1516 1517 1518 1519 1520 1521
{
    /* if there is a running dnsmasq, kill it */
    if (network->dnsmasqPid > 0) {
        networkKillDaemon(network->dnsmasqPid, "dnsmasq",
                          network->def->name);
        network->dnsmasqPid = -1;
1522
    }
1523
    /* now start dnsmasq if it should be started */
1524
    return networkStartDhcpDaemon(driver, network);
1525 1526
}

G
Gene Czarcinski 已提交
1527 1528 1529 1530 1531 1532
static char radvd1[] = "  AdvOtherConfigFlag off;\n\n";
static char radvd2[] = "    AdvAutonomous off;\n";
static char radvd3[] = "    AdvOnLink on;\n"
                       "    AdvAutonomous on;\n"
                       "    AdvRouterAddr off;\n";

1533 1534 1535
static int
networkRadvdConfContents(virNetworkObjPtr network, char **configstr)
{
E
Eric Blake 已提交
1536
    virBuffer configbuf = VIR_BUFFER_INITIALIZER;
1537 1538
    int ret = -1;
    size_t i;
1539
    virNetworkIPDefPtr ipdef;
G
Gene Czarcinski 已提交
1540
    bool v6present = false, dhcp6 = false;
1541 1542

    *configstr = NULL;
1543

G
Gene Czarcinski 已提交
1544
    /* Check if DHCPv6 is needed */
1545
    for (i = 0;
1546
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET6, i));
1547
         i++) {
G
Gene Czarcinski 已提交
1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560
        v6present = true;
        if (ipdef->nranges || ipdef->nhosts) {
            dhcp6 = true;
            break;
        }
    }

    /* If there are no IPv6 addresses, then we are done */
    if (!v6present) {
        ret = 0;
        goto cleanup;
    }

1561 1562 1563
    /* create radvd config file appropriate for this network;
     * IgnoreIfMissing allows radvd to start even when the bridge is down
     */
1564
    virBufferAsprintf(&configbuf, "interface %s\n"
1565 1566
                      "{\n"
                      "  AdvSendAdvert on;\n"
1567
                      "  IgnoreIfMissing on;\n"
G
Gene Czarcinski 已提交
1568 1569 1570 1571 1572
                      "  AdvManagedFlag %s;\n"
                      "%s",
                      network->def->bridge,
                      dhcp6 ? "on" : "off",
                      dhcp6 ? "\n" : radvd1);
1573 1574

    /* add a section for each IPv6 address in the config */
1575
    for (i = 0;
1576
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET6, i));
1577
         i++) {
1578 1579 1580
        int prefix;
        char *netaddr;

1581
        prefix = virNetworkIPDefPrefix(ipdef);
1582
        if (prefix < 0) {
1583 1584 1585
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge '%s' has an invalid prefix"),
                           network->def->bridge);
1586 1587
            goto cleanup;
        }
1588
        if (!(netaddr = virSocketAddrFormat(&ipdef->address)))
1589
            goto cleanup;
1590
        virBufferAsprintf(&configbuf,
1591
                          "  prefix %s/%d\n"
G
Gene Czarcinski 已提交
1592 1593 1594
                          "  {\n%s  };\n",
                          netaddr, prefix,
                          dhcp6 ? radvd2 : radvd3);
1595 1596 1597
        VIR_FREE(netaddr);
    }

1598
    virBufferAddLit(&configbuf, "};\n");
1599

1600
    if (virBufferCheckError(&configbuf) < 0)
1601
        goto cleanup;
1602

1603 1604
    *configstr = virBufferContentAndReset(&configbuf);

1605
    ret = 0;
1606
 cleanup:
1607 1608 1609 1610
    virBufferFreeAndReset(&configbuf);
    return ret;
}

1611
/* write file and return its name (which must be freed by caller) */
1612
static int
1613 1614 1615
networkRadvdConfWrite(virNetworkDriverStatePtr driver,
                      virNetworkObjPtr network,
                      char **configFile)
1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630
{
    int ret = -1;
    char *configStr = NULL;
    char *myConfigFile = NULL;

    if (!configFile)
        configFile = &myConfigFile;

    *configFile = NULL;

    if (networkRadvdConfContents(network, &configStr) < 0)
        goto cleanup;

    if (!configStr) {
        ret = 0;
1631 1632 1633 1634
        goto cleanup;
    }

    /* construct the filename */
1635
    if (!(*configFile = networkRadvdConfigFileName(driver, network->def->name)))
1636 1637
        goto cleanup;
    /* write the file */
1638
    if (virFileWriteStr(*configFile, configStr, 0600) < 0) {
1639 1640
        virReportSystemError(errno,
                             _("couldn't write radvd config file '%s'"),
1641 1642 1643 1644 1645
                             *configFile);
        goto cleanup;
    }

    ret = 0;
1646
 cleanup:
1647 1648 1649 1650 1651 1652
    VIR_FREE(configStr);
    VIR_FREE(myConfigFile);
    return ret;
}

static int
1653 1654
networkStartRadvd(virNetworkDriverStatePtr driver,
                  virNetworkObjPtr network)
1655
{
1656
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
1657 1658 1659 1660 1661 1662 1663 1664
    char *pidfile = NULL;
    char *radvdpidbase = NULL;
    char *configfile = NULL;
    virCommandPtr cmd = NULL;
    int ret = -1;

    network->radvdPid = -1;

G
Gene Czarcinski 已提交
1665
    /* Is dnsmasq handling RA? */
1666
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
G
Gene Czarcinski 已提交
1667 1668 1669 1670
        ret = 0;
        goto cleanup;
    }

1671
    if (!virNetworkDefGetIPByIndex(network->def, AF_INET6, 0)) {
1672 1673 1674 1675 1676 1677 1678 1679 1680 1681
        /* no IPv6 addresses, so we don't need to run radvd */
        ret = 0;
        goto cleanup;
    }

    if (!virFileIsExecutable(RADVD)) {
        virReportSystemError(errno,
                             _("Cannot find %s - "
                               "Possibly the package isn't installed"),
                             RADVD);
1682 1683 1684
        goto cleanup;
    }

1685
    if (virFileMakePath(driver->pidDir) < 0) {
1686 1687
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1688
                             driver->pidDir);
1689 1690
        goto cleanup;
    }
1691
    if (virFileMakePath(driver->radvdStateDir) < 0) {
1692 1693
        virReportSystemError(errno,
                             _("cannot create directory %s"),
1694
                             driver->radvdStateDir);
1695 1696 1697 1698
        goto cleanup;
    }

    /* construct pidfile name */
1699
    if (!(radvdpidbase = networkRadvdPidfileBasename(network->def->name)))
1700
        goto cleanup;
1701
    if (!(pidfile = virPidFileBuildPath(driver->pidDir, radvdpidbase)))
1702 1703
        goto cleanup;

1704
    if (networkRadvdConfWrite(driver, network, &configfile) < 0)
1705 1706
        goto cleanup;

1707 1708 1709 1710
    /* prevent radvd from daemonizing itself with "--debug 1", and use
     * a dummy pidfile name - virCommand will create the pidfile we
     * want to use (this is necessary because radvd's internal
     * daemonization and pidfile creation causes a race, and the
1711
     * virPidFileRead() below will fail if we use them).
1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726
     * Unfortunately, it isn't possible to tell radvd to not create
     * its own pidfile, so we just let it do so, with a slightly
     * different name. Unused, but harmless.
     */
    cmd = virCommandNewArgList(RADVD, "--debug", "1",
                               "--config", configfile,
                               "--pidfile", NULL);
    virCommandAddArgFormat(cmd, "%s-bin", pidfile);

    virCommandSetPidFile(cmd, pidfile);
    virCommandDaemonize(cmd);

    if (virCommandRun(cmd, NULL) < 0)
        goto cleanup;

1727
    if (virPidFileRead(driver->pidDir, radvdpidbase, &network->radvdPid) < 0)
1728 1729 1730
        goto cleanup;

    ret = 0;
1731
 cleanup:
1732
    virObjectUnref(dnsmasq_caps);
1733 1734 1735 1736 1737 1738 1739
    virCommandFree(cmd);
    VIR_FREE(configfile);
    VIR_FREE(radvdpidbase);
    VIR_FREE(pidfile);
    return ret;
}

1740
static int
1741 1742
networkRefreshRadvd(virNetworkDriverStatePtr driver,
                    virNetworkObjPtr network)
1743
{
1744
    dnsmasqCapsPtr dnsmasq_caps = networkGetDnsmasqCaps(driver);
G
Gene Czarcinski 已提交
1745 1746 1747
    char *radvdpidbase;

    /* Is dnsmasq handling RA? */
1748 1749
    if (DNSMASQ_RA_SUPPORT(dnsmasq_caps)) {
        virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
1750 1751 1752 1753 1754 1755 1756
        if (network->radvdPid <= 0)
            return 0;
        /* radvd should not be running but in case it is */
        if ((networkKillDaemon(network->radvdPid, "radvd",
                               network->def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))
             != NULL)) {
1757
            virPidFileDelete(driver->pidDir, radvdpidbase);
G
Gene Czarcinski 已提交
1758 1759 1760 1761 1762
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
        return 0;
    }
1763
    virObjectUnref(dnsmasq_caps);
G
Gene Czarcinski 已提交
1764

1765 1766
    /* if there's no running radvd, just start it */
    if (network->radvdPid <= 0 || (kill(network->radvdPid, 0) < 0))
1767
        return networkStartRadvd(driver, network);
1768

1769
    if (!virNetworkDefGetIPByIndex(network->def, AF_INET6, 0)) {
1770 1771 1772 1773
        /* no IPv6 addresses, so we don't need to run radvd */
        return 0;
    }

1774
    if (networkRadvdConfWrite(driver, network, NULL) < 0)
1775 1776 1777 1778 1779
        return -1;

    return kill(network->radvdPid, SIGHUP);
}

1780 1781
#if 0
/* currently unused, so it causes a build error unless we #if it out */
1782
static int
1783
networkRestartRadvd(virNetworkObjPtr network)
1784 1785 1786 1787 1788 1789 1790 1791 1792
{
    char *radvdpidbase;

    /* if there is a running radvd, kill it */
    if (network->radvdPid > 0) {
        /* essentially ignore errors from the following two functions,
         * since there's really no better recovery to be done than to
         * just push ahead (and that may be exactly what's needed).
         */
G
Gene Czarcinski 已提交
1793
        if ((networkKillDaemon(network->radvdPid, "radvd",
1794 1795 1796
                               network->def->name) >= 0) &&
            ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))
             != NULL)) {
1797
            virPidFileDelete(driver->pidDir, radvdpidbase);
1798 1799 1800 1801 1802 1803 1804 1805 1806
            VIR_FREE(radvdpidbase);
        }
        network->radvdPid = -1;
    }
    /* now start radvd if it should be started */
    return networkStartRadvd(network);
}
#endif /* #if 0 */

1807 1808
static int
networkRefreshDaemonsHelper(virNetworkObjPtr net,
1809
                            void *opaque)
1810
{
1811
    virNetworkDriverStatePtr driver = opaque;
1812

1813
    virObjectLock(net);
1814 1815 1816 1817 1818 1819 1820 1821 1822 1823
    if (virNetworkObjIsActive(net) &&
        ((net->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
        /* Only the three L3 network types that are configured by
         * libvirt will have a dnsmasq or radvd daemon associated
         * with them.  Here we send a SIGHUP to an existing
         * dnsmasq and/or radvd, or restart them if they've
         * disappeared.
         */
1824 1825
        networkRefreshDhcpDaemon(driver, net);
        networkRefreshRadvd(driver, net);
1826
    }
1827
    virObjectUnlock(net);
1828 1829 1830
    return 0;
}

1831 1832 1833 1834
/* SIGHUP/restart any dnsmasq or radvd daemons.
 * This should be called when libvirtd is restarted.
 */
static void
1835
networkRefreshDaemons(virNetworkDriverStatePtr driver)
1836 1837
{
    VIR_INFO("Refreshing network daemons");
1838 1839
    virNetworkObjListForEach(driver->networks,
                             networkRefreshDaemonsHelper,
1840
                             driver);
1841
}
1842

1843 1844 1845 1846 1847
static int
networkReloadFirewallRulesHelper(virNetworkObjPtr net,
                                 void *opaque ATTRIBUTE_UNUSED)
{

1848
    virObjectLock(net);
1849 1850 1851 1852 1853 1854 1855 1856 1857 1858
    if (virNetworkObjIsActive(net) &&
        ((net->def->forward.type == VIR_NETWORK_FORWARD_NONE) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_NAT) ||
         (net->def->forward.type == VIR_NETWORK_FORWARD_ROUTE))) {
        /* Only the three L3 network types that are configured by libvirt
         * need to have iptables rules reloaded.
         */
        networkRemoveFirewallRules(net->def);
        if (networkAddFirewallRules(net->def) < 0) {
            /* failed to add but already logged */
1859 1860
        }
    }
1861
    virObjectUnlock(net);
1862
    return 0;
1863 1864
}

1865
static void
1866
networkReloadFirewallRules(virNetworkDriverStatePtr driver)
1867
{
1868
    VIR_INFO("Reloading iptables rules");
1869 1870 1871
    virNetworkObjListForEach(driver->networks,
                             networkReloadFirewallRulesHelper,
                             NULL);
1872 1873
}

1874
/* Enable IP Forwarding. Return 0 for success, -1 for failure. */
1875
static int
1876
networkEnableIPForwarding(bool enableIPv4, bool enableIPv6)
1877
{
1878
    int ret = 0;
1879 1880 1881 1882
#ifdef HAVE_SYSCTLBYNAME
    int enabled = 1;
    if (enableIPv4)
        ret = sysctlbyname("net.inet.ip.forwarding", NULL, 0,
J
Ján Tomko 已提交
1883
                           &enabled, sizeof(enabled));
1884 1885
    if (enableIPv6 && ret == 0)
        ret = sysctlbyname("net.inet6.ip6.forwarding", NULL, 0,
J
Ján Tomko 已提交
1886
                           &enabled, sizeof(enabled));
1887
#else
1888 1889 1890 1891
    if (enableIPv4)
        ret = virFileWriteStr("/proc/sys/net/ipv4/ip_forward", "1\n", 0);
    if (enableIPv6 && ret == 0)
        ret = virFileWriteStr("/proc/sys/net/ipv6/conf/all/forwarding", "1\n", 0);
1892
#endif
1893
    return ret;
1894 1895
}

1896 1897
#define SYSCTL_PATH "/proc/sys"

1898 1899
static int
networkSetIPv6Sysctls(virNetworkObjPtr network)
1900 1901 1902
{
    char *field = NULL;
    int ret = -1;
1903
    bool enableIPv6 =  !!virNetworkDefGetIPByIndex(network->def, AF_INET6, 0);
1904

1905 1906 1907 1908 1909 1910 1911
    /* set disable_ipv6 if there are no ipv6 addresses defined for the
     * network. But also unset it if there *are* ipv6 addresses, as we
     * can't be sure of its default value.
     */
    if (virAsprintf(&field, SYSCTL_PATH "/net/ipv6/conf/%s/disable_ipv6",
                    network->def->bridge) < 0)
       goto cleanup;
1912

1913 1914
    if (access(field, W_OK) < 0 && errno == ENOENT) {
        if (!enableIPv6)
1915 1916
            VIR_DEBUG("ipv6 appears to already be disabled on %s",
                      network->def->bridge);
1917 1918 1919
        ret = 0;
        goto cleanup;
    }
1920

1921 1922 1923 1924 1925
    if (virFileWriteStr(field, enableIPv6 ? "0" : "1", 0) < 0) {
        virReportSystemError(errno,
                             _("cannot write to %s to enable/disable IPv6 "
                               "on bridge %s"), field, network->def->bridge);
        goto cleanup;
1926
    }
1927
    VIR_FREE(field);
1928

1929 1930
    /* The rest of the ipv6 sysctl tunables should always be set the
     * same, whether or not we're using ipv6 on this bridge.
1931 1932 1933 1934 1935 1936
     */

    /* Prevent guests from hijacking the host network by sending out
     * their own router advertisements.
     */
    if (virAsprintf(&field, SYSCTL_PATH "/net/ipv6/conf/%s/accept_ra",
1937
                    network->def->bridge) < 0)
1938 1939
        goto cleanup;

1940
    if (virFileWriteStr(field, "0", 0) < 0) {
1941
        virReportSystemError(errno,
1942 1943 1944 1945 1946
                             _("cannot disable %s"), field);
        goto cleanup;
    }
    VIR_FREE(field);

1947 1948 1949 1950
    /* All interfaces used as a gateway (which is what this is, by
     * definition), must always have autoconf=0.
     */
    if (virAsprintf(&field, SYSCTL_PATH "/net/ipv6/conf/%s/autoconf",
1951
                    network->def->bridge) < 0)
1952 1953
        goto cleanup;

1954
    if (virFileWriteStr(field, "0", 0) < 0) {
1955
        virReportSystemError(errno,
1956
                             _("cannot disable %s"), field);
1957 1958 1959 1960
        goto cleanup;
    }

    ret = 0;
1961
 cleanup:
1962 1963 1964 1965
    VIR_FREE(field);
    return ret;
}

1966
/* add an IP address to a bridge */
1967
static int
D
Daniel P. Berrange 已提交
1968
networkAddAddrToBridge(virNetworkObjPtr network,
1969
                       virNetworkIPDefPtr ipdef)
1970
{
1971
    int prefix = virNetworkIPDefPrefix(ipdef);
1972 1973

    if (prefix < 0) {
1974 1975 1976
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("bridge '%s' has an invalid netmask or IP address"),
                       network->def->bridge);
1977 1978 1979
        return -1;
    }

1980
    if (virNetDevSetIPAddress(network->def->bridge,
1981
                              &ipdef->address, NULL, prefix) < 0)
1982 1983 1984 1985 1986
        return -1;

    return 0;
}

1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

static int
networkStartHandleMACTableManagerMode(virNetworkObjPtr network,
                                      const char *macTapIfName)
{
    const char *brname = network->def->bridge;

    if (brname &&
        network->def->macTableManager
        == VIR_NETWORK_BRIDGE_MAC_TABLE_MANAGER_LIBVIRT) {
        if (virNetDevBridgeSetVlanFiltering(brname, true) < 0)
            return -1;
        if (macTapIfName) {
            if (virNetDevBridgePortSetLearning(brname, macTapIfName, false) < 0)
                return -1;
            if (virNetDevBridgePortSetUnicastFlood(brname, macTapIfName, false) < 0)
                return -1;
        }
    }
    return 0;
}


2010 2011 2012 2013 2014
/* add an IP (static) route to a bridge */
static int
networkAddRouteToBridge(virNetworkObjPtr network,
                        virNetworkRouteDefPtr routedef)
{
2015 2016 2017 2018
    int prefix = virNetworkRouteDefGetPrefix(routedef);
    unsigned int metric = virNetworkRouteDefGetMetric(routedef);
    virSocketAddrPtr addr = virNetworkRouteDefGetAddress(routedef);
    virSocketAddrPtr gateway = virNetworkRouteDefGetGateway(routedef);
2019 2020 2021 2022 2023 2024 2025 2026 2027

    if (prefix < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' has an invalid netmask "
                         "or IP address in route definition"),
                       network->def->name);
        return -1;
    }

2028 2029
    if (virNetDevAddRoute(network->def->bridge, addr,
                          prefix, gateway, metric) < 0) {
2030 2031 2032 2033 2034
        return -1;
    }
    return 0;
}

2035 2036 2037
static int
networkWaitDadFinish(virNetworkObjPtr network)
{
2038
    virNetworkIPDefPtr ipdef;
2039 2040 2041 2042 2043 2044
    virSocketAddrPtr *addrs = NULL, addr = NULL;
    size_t naddrs = 0;
    int ret = -1;

    VIR_DEBUG("Begin waiting for IPv6 DAD on network %s", network->def->name);

2045
    while ((ipdef = virNetworkDefGetIPByIndex(network->def,
2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060
                                              AF_INET6, naddrs))) {
        addr = &ipdef->address;
        if (VIR_APPEND_ELEMENT_COPY(addrs, naddrs, addr) < 0)
            goto cleanup;
    }

    ret = (naddrs == 0) ? 0 : virNetDevWaitDadFinish(addrs, naddrs);

 cleanup:
    VIR_FREE(addrs);
    VIR_DEBUG("Finished waiting for IPv6 DAD on network %s with status %d",
              network->def->name, ret);
    return ret;
}

2061
static int
2062 2063
networkStartNetworkVirtual(virNetworkDriverStatePtr driver,
                           virNetworkObjPtr network)
2064
{
2065
    size_t i;
2066
    bool v4present = false, v6present = false;
2067
    virErrorPtr save_err = NULL;
2068
    virNetworkIPDefPtr ipdef;
2069
    virNetworkRouteDefPtr routedef;
2070
    char *macTapIfName = NULL;
2071
    int tapfd = -1;
2072

2073
    /* Check to see if any network IP collides with an existing route */
2074
    if (networkCheckRouteCollision(network->def) < 0)
2075 2076
        return -1;

2077
    /* Create and configure the bridge device */
2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091
    if (!network->def->bridge) {
        /* bridge name can only be empty if the config files were
         * edited directly. Otherwise networkValidate() (called after
         * parsing the XML from networkCreateXML() and
         * networkDefine()) guarantees we will have a valid bridge
         * name before this point. Since hand editing of the config
         * files is explicitly prohibited we can, with clear
         * conscience, log an error and fail at this point.
         */
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' has no bridge name defined"),
                       network->def->name);
        return -1;
    }
2092
    if (virNetDevBridgeCreate(network->def->bridge) < 0)
2093 2094
        return -1;

2095 2096 2097 2098 2099 2100 2101 2102
    if (network->def->mac_specified) {
        /* To set a mac for the bridge, we need to define a dummy tap
         * device, set its mac, then attach it to the bridge. As long
         * as its mac address is lower than any other interface that
         * gets attached, the bridge will always maintain this mac
         * address.
         */
        macTapIfName = networkBridgeDummyNicName(network->def->bridge);
2103
        if (!macTapIfName)
2104
            goto err0;
2105
        /* Keep tun fd open and interface up to allow for IPv6 DAD to happen */
2106
        if (virNetDevTapCreateInBridgePort(network->def->bridge,
2107
                                           &macTapIfName, &network->def->mac,
2108
                                           NULL, NULL, &tapfd, 1, NULL, NULL,
2109 2110 2111
                                           VIR_NETDEV_TAP_CREATE_USE_MAC_FOR_BRIDGE |
                                           VIR_NETDEV_TAP_CREATE_IFUP |
                                           VIR_NETDEV_TAP_CREATE_PERSIST) < 0) {
2112 2113 2114 2115 2116
            VIR_FREE(macTapIfName);
            goto err0;
        }
    }

2117
    /* Set bridge options */
2118 2119 2120 2121

    /* delay is configured in seconds, but virNetDevBridgeSetSTPDelay
     * expects milliseconds
     */
2122
    if (virNetDevBridgeSetSTPDelay(network->def->bridge,
2123
                                   network->def->delay * 1000) < 0)
2124
        goto err1;
2125

2126
    if (virNetDevBridgeSetSTP(network->def->bridge,
2127
                              network->def->stp ? true : false) < 0)
2128
        goto err1;
2129

2130 2131 2132 2133
    /* Disable IPv6 on the bridge if there are no IPv6 addresses
     * defined, and set other IPv6 sysctl tunables appropriately.
     */
    if (networkSetIPv6Sysctls(network) < 0)
2134
        goto err1;
2135

2136
    /* Add "once per network" rules */
2137
    if (networkAddFirewallRules(network->def) < 0)
2138 2139
        goto err1;

2140
    for (i = 0;
2141
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_UNSPEC, i));
2142
         i++) {
2143
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET))
2144
            v4present = true;
2145
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6))
2146
            v6present = true;
2147

2148
        /* Add the IP address/netmask to the bridge */
2149
        if (networkAddAddrToBridge(network, ipdef) < 0)
2150
            goto err2;
2151 2152
    }

2153 2154 2155
    if (networkStartHandleMACTableManagerMode(network, macTapIfName) < 0)
        goto err2;

2156
    /* Bring up the bridge interface */
2157
    if (virNetDevSetOnline(network->def->bridge, 1) < 0)
2158
        goto err2;
2159

2160
    for (i = 0; i < network->def->nroutes; i++) {
2161 2162 2163 2164 2165
        virSocketAddrPtr gateway = NULL;

        routedef = network->def->routes[i];
        gateway = virNetworkRouteDefGetGateway(routedef);

2166 2167 2168
        /* Add the IP route to the bridge */
        /* ignore errors, error msg will be generated */
        /* but libvirt will not know and net-destroy will work. */
2169
        if (VIR_SOCKET_ADDR_VALID(gateway)) {
2170 2171 2172 2173 2174 2175 2176
            if (networkAddRouteToBridge(network, routedef) < 0) {
                /* an error occurred adding the static route */
                continue; /* for now, do nothing */
            }
        }
    }

2177 2178
    /* If forward.type != NONE, turn on global IP forwarding */
    if (network->def->forward.type != VIR_NETWORK_FORWARD_NONE &&
2179
        networkEnableIPForwarding(v4present, v6present) < 0) {
2180
        virReportSystemError(errno, "%s",
2181
                             _("failed to enable IP forwarding"));
2182
        goto err3;
2183 2184
    }

2185

2186
    /* start dnsmasq if there are any IP addresses (v4 or v6) */
2187
    if ((v4present || v6present) &&
2188
        networkStartDhcpDaemon(driver, network) < 0)
2189
        goto err3;
2190

2191
    /* start radvd if there are any ipv6 addresses */
2192
    if (v6present && networkStartRadvd(driver, network) < 0)
2193 2194
        goto err4;

2195 2196 2197 2198 2199 2200 2201 2202
    /* dnsmasq does not wait for DAD to complete before daemonizing,
     * so we need to wait for it ourselves.
     */
    if (v6present && networkWaitDadFinish(network) < 0)
        goto err4;

    /* DAD has finished, dnsmasq is now bound to the
     * bridge's IPv6 address, so we can set the dummy tun down.
2203 2204 2205 2206 2207 2208 2209
     */
    if (tapfd >= 0) {
        if (virNetDevSetOnline(macTapIfName, false) < 0)
            goto err4;
        VIR_FORCE_CLOSE(tapfd);
    }

2210
    if (virNetDevBandwidthSet(network->def->bridge,
2211
                              network->def->bandwidth, true) < 0)
2212 2213
        goto err5;

2214
    VIR_FREE(macTapIfName);
2215 2216 2217

    return 0;

2218
 err5:
2219 2220
    if (network->def->bandwidth)
       virNetDevBandwidthClear(network->def->bridge);
2221

2222 2223 2224 2225
 err4:
    if (!save_err)
        save_err = virSaveLastError();

2226 2227 2228 2229 2230
    if (network->dnsmasqPid > 0) {
        kill(network->dnsmasqPid, SIGTERM);
        network->dnsmasqPid = -1;
    }

2231 2232 2233
 err3:
    if (!save_err)
        save_err = virSaveLastError();
2234
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2235

2236 2237 2238
 err2:
    if (!save_err)
        save_err = virSaveLastError();
2239
    networkRemoveFirewallRules(network->def);
2240 2241

 err1:
2242 2243 2244
    if (!save_err)
        save_err = virSaveLastError();

H
Hu Tao 已提交
2245
    if (macTapIfName) {
2246
        VIR_FORCE_CLOSE(tapfd);
2247
        ignore_value(virNetDevTapDelete(macTapIfName, NULL));
H
Hu Tao 已提交
2248 2249
        VIR_FREE(macTapIfName);
    }
2250 2251

 err0:
2252 2253
    if (!save_err)
        save_err = virSaveLastError();
2254
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2255

2256 2257 2258 2259
    if (save_err) {
        virSetError(save_err);
        virFreeError(save_err);
    }
2260
    /* coverity[leaked_handle] - 'tapfd' is not leaked */
2261 2262 2263
    return -1;
}

2264 2265 2266
static int
networkShutdownNetworkVirtual(virNetworkDriverStatePtr driver,
                              virNetworkObjPtr network)
2267
{
2268 2269
    if (network->def->bandwidth)
        virNetDevBandwidthClear(network->def->bridge);
2270

2271 2272 2273 2274 2275
    if (network->radvdPid > 0) {
        char *radvdpidbase;

        kill(network->radvdPid, SIGTERM);
        /* attempt to delete the pidfile we created */
2276
        if ((radvdpidbase = networkRadvdPidfileBasename(network->def->name))) {
2277
            virPidFileDelete(driver->pidDir, radvdpidbase);
2278 2279 2280 2281
            VIR_FREE(radvdpidbase);
        }
    }

2282 2283 2284
    if (network->dnsmasqPid > 0)
        kill(network->dnsmasqPid, SIGTERM);

2285
    if (network->def->mac_specified) {
2286
        char *macTapIfName = networkBridgeDummyNicName(network->def->bridge);
2287
        if (macTapIfName) {
2288
            ignore_value(virNetDevTapDelete(macTapIfName, NULL));
2289 2290 2291 2292
            VIR_FREE(macTapIfName);
        }
    }

2293
    ignore_value(virNetDevSetOnline(network->def->bridge, 0));
2294

2295
    networkRemoveFirewallRules(network->def);
2296

2297
    ignore_value(virNetDevBridgeDelete(network->def->bridge));
2298

2299
    /* See if its still alive and really really kill it */
2300
    if (network->dnsmasqPid > 0 &&
2301
        (kill(network->dnsmasqPid, 0) == 0))
2302 2303
        kill(network->dnsmasqPid, SIGKILL);
    network->dnsmasqPid = -1;
2304 2305 2306 2307 2308 2309

    if (network->radvdPid > 0 &&
        (kill(network->radvdPid, 0) == 0))
        kill(network->radvdPid, SIGKILL);
    network->radvdPid = -1;

2310 2311 2312
    return 0;
}

2313

2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334
static int
networkStartNetworkBridge(virNetworkObjPtr network)
{
    /* put anything here that needs to be done each time a network of
     * type BRIDGE, is started. On failure, undo anything you've done,
     * and return -1. On success return 0.
     */
    return networkStartHandleMACTableManagerMode(network, NULL);
}

static int
networkShutdownNetworkBridge(virNetworkObjPtr network ATTRIBUTE_UNUSED)
{
    /* put anything here that needs to be done each time a network of
     * type BRIDGE is shutdown. On failure, undo anything you've done,
     * and return -1. On success return 0.
     */
    return 0;
}


2335 2336 2337 2338 2339 2340 2341 2342 2343
/* networkCreateInterfacePool:
 * @netdef: the original NetDef from the network
 *
 * Creates an implicit interface pool of VF's when a PF dev is given
 */
static int
networkCreateInterfacePool(virNetworkDefPtr netdef)
{
    size_t numVirtFns = 0;
2344
    unsigned int maxVirtFns = 0;
2345 2346 2347 2348 2349 2350
    char **vfNames = NULL;
    virPCIDeviceAddressPtr *virtFns;

    int ret = -1;
    size_t i;

2351 2352 2353
    if (netdef->forward.npfs == 0 || netdef->forward.nifs > 0)
       return 0;

2354 2355
    if ((virNetDevGetVirtualFunctions(netdef->forward.pfs->dev, &vfNames,
                                      &virtFns, &numVirtFns, &maxVirtFns)) < 0) {
2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Could not get Virtual functions on %s"),
                       netdef->forward.pfs->dev);
        goto cleanup;
    }

    if (VIR_ALLOC_N(netdef->forward.ifs, numVirtFns) < 0)
        goto cleanup;

    for (i = 0; i < numVirtFns; i++) {
        virPCIDeviceAddressPtr thisVirtFn = virtFns[i];
        const char *thisName = vfNames[i];
        virNetworkForwardIfDefPtr thisIf
            = &netdef->forward.ifs[netdef->forward.nifs];

        switch (netdef->forward.type) {
        case VIR_NETWORK_FORWARD_BRIDGE:
        case VIR_NETWORK_FORWARD_PRIVATE:
        case VIR_NETWORK_FORWARD_VEPA:
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
            if (thisName) {
                if (VIR_STRDUP(thisIf->device.dev, thisName) < 0)
                    goto cleanup;
                thisIf->type = VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV;
                netdef->forward.nifs++;
            } else {
                VIR_WARN("VF %zu of SRIOV PF %s couldn't be added to the "
                         "interface pool because it isn't bound "
                         "to a network driver - possibly in use elsewhere",
                         i, netdef->forward.pfs->dev);
            }
            break;

        case VIR_NETWORK_FORWARD_HOSTDEV:
            /* VF's are always PCI devices */
            thisIf->type = VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI;
            thisIf->device.pci.domain = thisVirtFn->domain;
            thisIf->device.pci.bus = thisVirtFn->bus;
            thisIf->device.pci.slot = thisVirtFn->slot;
            thisIf->device.pci.function = thisVirtFn->function;
            netdef->forward.nifs++;
            break;

        case VIR_NETWORK_FORWARD_NONE:
        case VIR_NETWORK_FORWARD_NAT:
        case VIR_NETWORK_FORWARD_ROUTE:
        case VIR_NETWORK_FORWARD_LAST:
            /* by definition these will never be encountered here */
            break;
        }
    }

    if (netdef->forward.nifs == 0) {
        /* If we don't get at least one interface in the pool, declare
         * failure
         */
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("No usable Vf's present on SRIOV PF %s"),
                       netdef->forward.pfs->dev);
        goto cleanup;
    }

    ret = 0;
 cleanup:
    if (ret < 0) {
        /* free all the entries made before error */
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV)
                VIR_FREE(netdef->forward.ifs[i].device.dev);
        }
        netdef->forward.nifs = 0;
    }
    if (netdef->forward.nifs == 0)
        VIR_FREE(netdef->forward.ifs);

    for (i = 0; i < numVirtFns; i++) {
        VIR_FREE(vfNames[i]);
        VIR_FREE(virtFns[i]);
    }
    VIR_FREE(vfNames);
    VIR_FREE(virtFns);
    return ret;
}


2442
static int
2443
networkStartNetworkExternal(virNetworkObjPtr network)
2444 2445
{
    /* put anything here that needs to be done each time a network of
2446
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is started. On
2447 2448 2449
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
2450
    return networkCreateInterfacePool(network->def);
2451 2452
}

2453
static int networkShutdownNetworkExternal(virNetworkObjPtr network ATTRIBUTE_UNUSED)
2454 2455
{
    /* put anything here that needs to be done each time a network of
2456
     * type BRIDGE, PRIVATE, VEPA, HOSTDEV or PASSTHROUGH is shutdown. On
2457 2458 2459 2460 2461 2462 2463
     * failure, undo anything you've done, and return -1. On success
     * return 0.
     */
    return 0;
}

static int
2464 2465
networkStartNetwork(virNetworkDriverStatePtr driver,
                    virNetworkObjPtr network)
2466
{
2467 2468 2469
    int ret = -1;

    VIR_DEBUG("driver=%p, network=%p", driver, network);
2470 2471

    if (virNetworkObjIsActive(network)) {
2472 2473
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("network is already active"));
2474
        return ret;
2475 2476
    }

2477 2478 2479
    VIR_DEBUG("Beginning network startup process");

    VIR_DEBUG("Setting current network def as transient");
2480
    if (virNetworkObjSetDefTransient(network, true) < 0)
2481
        goto cleanup;
2482

2483 2484
    /* Run an early hook to set-up missing devices.
     * If the script raised an error abort the launch. */
2485
    if (networkRunHook(network, NULL, NULL,
2486 2487 2488 2489
                       VIR_HOOK_NETWORK_OP_START,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2490
    switch (network->def->forward.type) {
2491 2492 2493 2494

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2495
        if (networkStartNetworkVirtual(driver, network) < 0)
2496
            goto cleanup;
2497 2498 2499
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2500 2501 2502 2503 2504 2505 2506 2507 2508
        if (network->def->bridge) {
            if (networkStartNetworkBridge(network) < 0)
                goto cleanup;
            break;
        }
        /* intentionally fall through to the macvtap/direct case for
         * VIR_NETWORK_FORWARD_BRIDGE with no bridge device defined
         * (since that is macvtap bridge mode).
         */
2509 2510 2511
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2512
    case VIR_NETWORK_FORWARD_HOSTDEV:
2513
        if (networkStartNetworkExternal(network) < 0)
2514
            goto cleanup;
2515 2516 2517
        break;
    }

2518
    /* finally we can call the 'started' hook script if any */
2519
    if (networkRunHook(network, NULL, NULL,
2520 2521 2522 2523
                       VIR_HOOK_NETWORK_OP_STARTED,
                       VIR_HOOK_SUBOP_BEGIN) < 0)
        goto cleanup;

2524 2525 2526
    /* Persist the live configuration now that anything autogenerated
     * is setup.
     */
2527
    VIR_DEBUG("Writing network status to disk");
2528
    if (virNetworkSaveStatus(driver->stateDir, network) < 0)
2529
        goto cleanup;
2530 2531

    network->active = 1;
2532 2533
    VIR_INFO("Network '%s' started up", network->def->name);
    ret = 0;
2534

2535
 cleanup:
2536
    if (ret < 0) {
2537
        virNetworkObjUnsetDefTransient(network);
2538 2539
        virErrorPtr save_err = virSaveLastError();
        int save_errno = errno;
2540
        networkShutdownNetwork(driver, network);
2541 2542 2543 2544 2545 2546 2547
        virSetError(save_err);
        virFreeError(save_err);
        errno = save_errno;
    }
    return ret;
}

2548 2549 2550
static int
networkShutdownNetwork(virNetworkDriverStatePtr driver,
                       virNetworkObjPtr network)
2551 2552 2553 2554 2555 2556 2557 2558 2559
{
    int ret = 0;
    char *stateFile;

    VIR_INFO("Shutting down network '%s'", network->def->name);

    if (!virNetworkObjIsActive(network))
        return 0;

2560
    stateFile = virNetworkConfigFile(driver->stateDir,
2561
                                     network->def->name);
2562 2563 2564 2565 2566 2567
    if (!stateFile)
        return -1;

    unlink(stateFile);
    VIR_FREE(stateFile);

2568
    switch (network->def->forward.type) {
2569 2570 2571 2572

    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
2573
        ret = networkShutdownNetworkVirtual(driver, network);
2574 2575 2576
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
2577 2578 2579 2580 2581 2582 2583 2584
        if (network->def->bridge) {
            ret = networkShutdownNetworkBridge(network);
            break;
        }
        /* intentionally fall through to the macvtap/direct case for
         * VIR_NETWORK_FORWARD_BRIDGE with no bridge device defined
         * (since that is macvtap bridge mode).
         */
2585 2586 2587
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
2588
    case VIR_NETWORK_FORWARD_HOSTDEV:
2589
        ret = networkShutdownNetworkExternal(network);
2590 2591 2592
        break;
    }

2593
    /* now that we know it's stopped call the hook if present */
2594
    networkRunHook(network, NULL, NULL, VIR_HOOK_NETWORK_OP_STOPPED,
2595 2596
                   VIR_HOOK_SUBOP_END);

2597
    network->active = 0;
2598
    virNetworkObjUnsetDefTransient(network);
2599
    return ret;
2600 2601 2602
}


2603
static virNetworkPtr networkLookupByUUID(virConnectPtr conn,
2604 2605
                                         const unsigned char *uuid)
{
2606
    virNetworkDriverStatePtr driver = networkGetDriver();
2607 2608
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;
2609

2610
    network = virNetworkObjFindByUUID(driver->networks, uuid);
2611
    if (!network) {
2612 2613
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(uuid, uuidstr);
2614
        virReportError(VIR_ERR_NO_NETWORK,
2615 2616
                       _("no network with matching uuid '%s'"),
                       uuidstr);
2617
        goto cleanup;
2618 2619
    }

2620 2621 2622
    if (virNetworkLookupByUUIDEnsureACL(conn, network->def) < 0)
        goto cleanup;

2623 2624
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2625
 cleanup:
2626
    virNetworkObjEndAPI(&network);
2627
    return ret;
2628 2629
}

2630
static virNetworkPtr networkLookupByName(virConnectPtr conn,
2631 2632
                                         const char *name)
{
2633
    virNetworkDriverStatePtr driver = networkGetDriver();
2634 2635 2636
    virNetworkObjPtr network;
    virNetworkPtr ret = NULL;

2637
    network = virNetworkObjFindByName(driver->networks, name);
2638
    if (!network) {
2639 2640
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"), name);
2641
        goto cleanup;
2642 2643
    }

2644 2645 2646
    if (virNetworkLookupByNameEnsureACL(conn, network->def) < 0)
        goto cleanup;

2647 2648
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

2649
 cleanup:
2650
    virNetworkObjEndAPI(&network);
2651
    return ret;
2652 2653
}

2654 2655
static int networkConnectNumOfNetworks(virConnectPtr conn)
{
2656
    virNetworkDriverStatePtr driver = networkGetDriver();
2657
    int nactive;
2658

2659 2660 2661
    if (virConnectNumOfNetworksEnsureACL(conn) < 0)
        return -1;

2662 2663 2664 2665
    nactive = virNetworkObjListNumOfNetworks(driver->networks,
                                             true,
                                             virConnectNumOfNetworksCheckACL,
                                             conn);
2666

2667 2668 2669
    return nactive;
}

2670 2671 2672 2673 2674
static int networkConnectListNetworks(virConnectPtr conn,
                                      char **const names,
                                      int nnames)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
2675
    int got = 0;
2676

2677 2678 2679
    if (virConnectListNetworksEnsureACL(conn) < 0)
        return -1;

2680 2681 2682 2683
    got = virNetworkObjListGetNames(driver->networks,
                                    true, names, nnames,
                                    virConnectListNetworksCheckACL,
                                    conn);
2684

2685 2686 2687
    return got;
}

2688 2689
static int networkConnectNumOfDefinedNetworks(virConnectPtr conn)
{
2690
    virNetworkDriverStatePtr driver = networkGetDriver();
2691
    int ninactive = 0;
2692

2693 2694 2695
    if (virConnectNumOfDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2696 2697 2698 2699
    ninactive = virNetworkObjListNumOfNetworks(driver->networks,
                                               false,
                                               virConnectNumOfDefinedNetworksCheckACL,
                                               conn);
2700

2701 2702 2703
    return ninactive;
}

2704 2705 2706
static int networkConnectListDefinedNetworks(virConnectPtr conn, char **const names, int nnames)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
2707
    int got = 0;
2708

2709 2710 2711
    if (virConnectListDefinedNetworksEnsureACL(conn) < 0)
        return -1;

2712 2713 2714 2715
    got = virNetworkObjListGetNames(driver->networks,
                                    false, names, nnames,
                                    virConnectListDefinedNetworksCheckACL,
                                    conn);
2716 2717 2718
    return got;
}

2719
static int
2720 2721 2722
networkConnectListAllNetworks(virConnectPtr conn,
                              virNetworkPtr **nets,
                              unsigned int flags)
2723
{
2724
    virNetworkDriverStatePtr driver = networkGetDriver();
2725 2726 2727 2728
    int ret = -1;

    virCheckFlags(VIR_CONNECT_LIST_NETWORKS_FILTERS_ALL, -1);

2729 2730 2731
    if (virConnectListAllNetworksEnsureACL(conn) < 0)
        goto cleanup;

2732
    ret = virNetworkObjListExport(conn, driver->networks, nets,
2733 2734
                                  virConnectListAllNetworksCheckACL,
                                  flags);
2735

2736
 cleanup:
2737 2738
    return ret;
}
2739

2740 2741 2742 2743 2744 2745 2746 2747
static int
networkConnectNetworkEventRegisterAny(virConnectPtr conn,
                                      virNetworkPtr net,
                                      int eventID,
                                      virConnectNetworkEventGenericCallback callback,
                                      void *opaque,
                                      virFreeCallback freecb)
{
2748
    virNetworkDriverStatePtr driver = networkGetDriver();
2749 2750 2751 2752 2753 2754
    int ret = -1;

    if (virConnectNetworkEventRegisterAnyEnsureACL(conn) < 0)
        goto cleanup;

    if (virNetworkEventStateRegisterID(conn, driver->networkEventState,
2755
                                       net, eventID, callback,
2756 2757 2758
                                       opaque, freecb, &ret) < 0)
        ret = -1;

2759
 cleanup:
2760 2761 2762 2763 2764 2765 2766
    return ret;
}

static int
networkConnectNetworkEventDeregisterAny(virConnectPtr conn,
                                        int callbackID)
{
2767
    virNetworkDriverStatePtr driver = networkGetDriver();
2768 2769 2770 2771 2772
    int ret = -1;

    if (virConnectNetworkEventDeregisterAnyEnsureACL(conn) < 0)
        goto cleanup;

2773 2774 2775 2776 2777 2778
    if (virObjectEventStateDeregisterID(conn,
                                        driver->networkEventState,
                                        callbackID) < 0)
        goto cleanup;

    ret = 0;
2779

2780
 cleanup:
2781 2782 2783
    return ret;
}

2784 2785 2786 2787 2788
static int networkIsActive(virNetworkPtr net)
{
    virNetworkObjPtr obj;
    int ret = -1;

2789 2790
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2791 2792 2793 2794

    if (virNetworkIsActiveEnsureACL(net->conn, obj->def) < 0)
        goto cleanup;

2795 2796
    ret = virNetworkObjIsActive(obj);

2797
 cleanup:
2798
    virNetworkObjEndAPI(&obj);
2799 2800 2801 2802 2803 2804 2805 2806
    return ret;
}

static int networkIsPersistent(virNetworkPtr net)
{
    virNetworkObjPtr obj;
    int ret = -1;

2807 2808
    if (!(obj = networkObjFromNetwork(net)))
        return ret;
2809 2810 2811 2812

    if (virNetworkIsPersistentEnsureACL(net->conn, obj->def) < 0)
        goto cleanup;

2813 2814
    ret = obj->persistent;

2815
 cleanup:
2816
    virNetworkObjEndAPI(&obj);
2817 2818 2819 2820
    return ret;
}


2821 2822
/*
 * networkFindUnusedBridgeName() - try to find a bridge name that is
2823 2824 2825
 * unused by the currently configured libvirt networks, as well as by
 * the host system itself (possibly created by someone/something other
 * than libvirt). Set this network's name to that new name.
2826 2827 2828 2829 2830 2831 2832 2833
 */
static int
networkFindUnusedBridgeName(virNetworkObjListPtr nets,
                            virNetworkDefPtr def)
{

    int ret = -1, id = 0;
    char *newname = NULL;
2834 2835 2836 2837 2838
    const char *templ = "virbr%d";
    const char *p;

    if (def->bridge &&
        (p = strchr(def->bridge, '%')) == strrchr(def->bridge, '%') &&
2839
        p && p[1] == 'd')
2840
        templ = def->bridge;
2841 2842 2843 2844

    do {
        if (virAsprintf(&newname, templ, id) < 0)
            goto cleanup;
2845 2846 2847 2848 2849 2850 2851
        /* check if this name is used in another libvirt network or
         * there is an existing device with that name. ignore errors
         * from virNetDevExists(), just in case it isn't implemented
         * on this platform (probably impossible).
         */
        if (!(virNetworkBridgeInUse(nets, newname, def->name) ||
              virNetDevExists(newname) == 1)) {
2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903
            VIR_FREE(def->bridge); /*could contain template */
            def->bridge = newname;
            ret = 0;
            goto cleanup;
        }
        VIR_FREE(newname);
    } while (++id <= MAX_BRIDGE_ID);

    virReportError(VIR_ERR_INTERNAL_ERROR,
                   _("Bridge generation exceeded max id %d"),
                   MAX_BRIDGE_ID);
    ret = 0;
 cleanup:
    if (ret < 0)
        VIR_FREE(newname);
    return ret;
}



/*
 * networkValidateBridgeName() - if no bridge name is set, or if the
 * bridge name contains a %d (indicating that this is a template for
 * the actual name) try to set an appropriate bridge name.  If a
 * bridge name *is* set, make sure it doesn't conflict with any other
 * network's bridge name.
 */
static int
networkBridgeNameValidate(virNetworkObjListPtr nets,
                          virNetworkDefPtr def)
{
    int ret = -1;

    if (def->bridge && !strstr(def->bridge, "%d")) {
        if (virNetworkBridgeInUse(nets, def->bridge, def->name)) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("bridge name '%s' already in use."),
                           def->bridge);
            goto cleanup;
        }
    } else {
        /* Allocate a bridge name */
        if (networkFindUnusedBridgeName(nets, def) < 0)
            goto cleanup;
    }

    ret = 0;
 cleanup:
    return ret;
}


2904
static int
2905
networkValidate(virNetworkDriverStatePtr driver,
2906
                virNetworkDefPtr def)
2907
{
2908
    size_t i, j;
2909 2910
    bool vlanUsed, vlanAllowed, badVlanUse = false;
    virPortGroupDefPtr defaultPortGroup = NULL;
2911
    virNetworkIPDefPtr ipdef;
G
Gene Czarcinski 已提交
2912
    bool ipv4def = false, ipv6def = false;
2913
    bool bandwidthAllowed = true;
2914
    bool usesInterface = false, usesAddress = false;
2915 2916 2917 2918

    /* Only the three L3 network types that are configured by libvirt
     * need to have a bridge device name / mac address provided
     */
2919 2920 2921
    if (def->forward.type == VIR_NETWORK_FORWARD_NONE ||
        def->forward.type == VIR_NETWORK_FORWARD_NAT ||
        def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
2922

2923 2924 2925 2926
        /* if no bridge name was given in the config, find a name
         * unused by any other libvirt networks and assign it.
         */
        if (networkBridgeNameValidate(driver->networks, def) < 0)
2927 2928 2929
            return -1;

        virNetworkSetBridgeMacAddr(def);
2930 2931
    } else {
        /* They are also the only types that currently support setting
2932 2933
         * a MAC or IP address for the host-side device (bridge), DNS
         * configuration, or network-wide bandwidth limits.
2934
         */
2935 2936 2937 2938 2939 2940 2941 2942
        if (def->mac_specified) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <mac> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
                           virNetworkForwardTypeToString(def->forward.type));
            return -1;
        }
2943
        if (virNetworkDefGetIPByIndex(def, AF_UNSPEC, 0)) {
2944 2945 2946 2947
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <ip> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2948
                           virNetworkForwardTypeToString(def->forward.type));
2949 2950
            return -1;
        }
2951
        if (def->dns.ntxts || def->dns.nhosts || def->dns.nsrvs) {
2952 2953 2954 2955
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <dns> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2956
                           virNetworkForwardTypeToString(def->forward.type));
2957 2958 2959 2960 2961 2962 2963
            return -1;
        }
        if (def->domain) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <domain> element in network %s "
                             "with forward mode='%s'"),
                           def->name,
2964
                           virNetworkForwardTypeToString(def->forward.type));
2965 2966
            return -1;
        }
2967 2968 2969 2970 2971 2972 2973 2974
        if (def->bandwidth) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported network-wide <bandwidth> element "
                             "in network %s with forward mode='%s'"),
                           def->name,
                           virNetworkForwardTypeToString(def->forward.type));
            return -1;
        }
2975
        bandwidthAllowed = false;
2976 2977
    }

2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011
    /* we support configs with a single PF defined:
     *   <pf dev='eth0'/>
     * or with a list of netdev names:
     *   <interface dev='eth9'/>
     * OR a list of PCI addresses
     *   <address type='pci' domain='0' bus='4' slot='0' function='1'/>
     * but not any combination of those.
     *
     * Since <interface> and <address> are for some strange reason
     * stored in the same array, we need to cycle through it and check
     * the type of each.
     */
    for (i = 0; i < def->forward.nifs; i++) {
        switch ((virNetworkForwardHostdevDeviceType)
                def->forward.ifs[i].type) {
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV:
            usesInterface = true;
            break;
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI:
            usesAddress = true;
            break;
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NONE:
        case VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_LAST:
            break;
        }
    }
    if ((def->forward.npfs > 0) + usesInterface + usesAddress > 1) {
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<address>, <interface>, and <pf> elements of "
                         "<forward> in network %s are mutually exclusive"),
                       def->name);
        return -1;
    }

G
Gene Czarcinski 已提交
3012 3013 3014
    /* We only support dhcp on one IPv4 address and
     * on one IPv6 address per defined network
     */
3015
    for (i = 0;
3016
         (ipdef = virNetworkDefGetIPByIndex(def, AF_UNSPEC, i));
3017
         i++) {
G
Gene Czarcinski 已提交
3018 3019 3020 3021 3022
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv4def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("Multiple IPv4 dhcp sections found -- "
3023 3024
                                 "dhcp is supported only for a "
                                 "single IPv4 address on each network"));
G
Gene Czarcinski 已提交
3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041
                    return -1;
                } else {
                    ipv4def = true;
                }
            }
        }
        if (VIR_SOCKET_ADDR_IS_FAMILY(&ipdef->address, AF_INET6)) {
            if (ipdef->nranges || ipdef->nhosts) {
                if (ipv6def) {
                    virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                               _("Multiple IPv6 dhcp sections found -- "
                                 "dhcp is supported only for a "
                                 "single IPv6 address on each network"));
                    return -1;
                } else {
                    ipv6def = true;
                }
3042 3043 3044
            }
        }
    }
3045 3046 3047 3048 3049 3050

    /* The only type of networks that currently support transparent
     * vlan configuration are those using hostdev sr-iov devices from
     * a pool, and those using an Open vSwitch bridge.
     */

3051 3052 3053
    vlanAllowed = (def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV ||
                   def->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH ||
                   (def->forward.type == VIR_NETWORK_FORWARD_BRIDGE &&
J
Ján Tomko 已提交
3054 3055
                    def->virtPortProfile &&
                    def->virtPortProfile->virtPortType
3056
                    == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH));
3057 3058

    vlanUsed = def->vlan.nTags > 0;
3059 3060
    for (i = 0; i < def->nPortGroups; i++) {
        if (vlanUsed || def->portGroups[i].vlan.nTags > 0) {
3061 3062 3063 3064 3065
            /* anyone using this portgroup will get a vlan tag. Verify
             * that they will also be using an openvswitch connection,
             * as that is the only type of network that currently
             * supports a vlan tag.
             */
3066
            if (def->portGroups[i].virtPortProfile) {
3067
                if (def->forward.type != VIR_NETWORK_FORWARD_BRIDGE ||
3068
                    def->portGroups[i].virtPortProfile->virtPortType
3069 3070 3071 3072 3073 3074 3075
                    != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                    badVlanUse = true;
                }
            } else if (!vlanAllowed) {
                /* virtualport taken from base network definition */
                badVlanUse = true;
            }
3076
        }
3077
        if (def->portGroups[i].isDefault) {
3078 3079 3080 3081 3082
            if (defaultPortGroup) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("network '%s' has multiple default "
                                 "<portgroup> elements (%s and %s), "
                                 "but only one default is allowed"),
3083
                               def->name, defaultPortGroup->name,
3084
                               def->portGroups[i].name);
3085
                return -1;
3086
            }
3087
            defaultPortGroup = &def->portGroups[i];
3088
        }
3089 3090 3091 3092 3093 3094 3095 3096 3097
        for (j = i + 1; j < def->nPortGroups; j++) {
            if (STREQ(def->portGroups[i].name, def->portGroups[j].name)) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("multiple <portgroup> elements with the "
                                 "same name (%s) in network '%s'"),
                               def->portGroups[i].name, def->name);
                return -1;
            }
        }
3098 3099 3100 3101 3102 3103 3104 3105
        if (def->portGroups[i].bandwidth && !bandwidthAllowed) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                           _("Unsupported <bandwidth> element in network '%s' "
                             "in portgroup '%s' with forward mode='%s'"),
                           def->name, def->portGroups[i].name,
                           virNetworkForwardTypeToString(def->forward.type));
            return -1;
        }
3106
    }
3107 3108 3109 3110 3111 3112 3113
    if (badVlanUse ||
        (vlanUsed && !vlanAllowed && !defaultPortGroup)) {
        /* NB: if defaultPortGroup is set, we don't directly look at
         * vlanUsed && !vlanAllowed, because the network will never be
         * used without having a portgroup added in, so all necessary
         * checks were done in the loop above.
         */
3114 3115 3116 3117 3118 3119
        virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                       _("<vlan> element specified for network %s, "
                         "whose type doesn't support vlan configuration"),
                       def->name);
        return -1;
    }
3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133

    if (def->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
        for (i = 0; i < def->nPortGroups; i++) {
            if (def->portGroups[i].bandwidth) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("unsupported <bandwidth> element "
                                 "in <portgroup name='%s'> of "
                                 "network '%s' with forward mode='%s'"),
                               def->portGroups[i].name, def->name,
                               virNetworkForwardTypeToString(def->forward.type));
                return -1;
            }
        }
    }
3134 3135 3136
    return 0;
}

3137 3138
static virNetworkPtr networkCreateXML(virConnectPtr conn, const char *xml)
{
3139
    virNetworkDriverStatePtr driver = networkGetDriver();
3140
    virNetworkDefPtr def;
3141
    virNetworkObjPtr network = NULL;
3142
    virNetworkPtr ret = NULL;
3143
    virObjectEventPtr event = NULL;
3144

3145
    if (!(def = virNetworkDefParseString(xml)))
3146
        goto cleanup;
3147

3148 3149 3150
    if (virNetworkCreateXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3151
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3152
        goto cleanup;
3153

3154 3155 3156
    /* NB: even though this transient network hasn't yet been started,
     * we assign the def with live = true in anticipation that it will
     * be started momentarily.
3157
     */
3158 3159 3160
    if (!(network = virNetworkAssignDef(driver->networks, def,
                                        VIR_NETWORK_OBJ_LIST_ADD_LIVE |
                                        VIR_NETWORK_OBJ_LIST_ADD_CHECK_LIVE)))
3161 3162
        goto cleanup;
    def = NULL;
3163

3164
    if (networkStartNetwork(driver, network) < 0) {
3165
        virNetworkRemoveInactive(driver->networks,
3166
                                 network);
3167
        goto cleanup;
3168 3169
    }

3170 3171
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3172 3173
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3174

3175
    VIR_INFO("Creating network '%s'", network->def->name);
3176 3177
    ret = virGetNetwork(conn, network->def->name, network->def->uuid);

3178
 cleanup:
3179
    virNetworkDefFree(def);
3180 3181
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3182
    virNetworkObjEndAPI(&network);
3183
    return ret;
3184 3185
}

3186 3187
static virNetworkPtr networkDefineXML(virConnectPtr conn, const char *xml)
{
3188
    virNetworkDriverStatePtr driver = networkGetDriver();
3189
    virNetworkDefPtr def = NULL;
3190
    bool freeDef = true;
3191
    virNetworkObjPtr network = NULL;
3192
    virNetworkPtr ret = NULL;
3193
    virObjectEventPtr event = NULL;
3194

3195
    if (!(def = virNetworkDefParseString(xml)))
3196
        goto cleanup;
3197

3198 3199 3200
    if (virNetworkDefineXMLEnsureACL(conn, def) < 0)
        goto cleanup;

3201
    if (networkValidate(driver, def) < 0)
J
Ján Tomko 已提交
3202
        goto cleanup;
3203

3204
    if (!(network = virNetworkAssignDef(driver->networks, def, 0)))
J
Ján Tomko 已提交
3205
        goto cleanup;
3206

3207
    /* def was assigned to network object */
3208
    freeDef = false;
3209 3210

    if (virNetworkSaveConfig(driver->networkConfigDir, def) < 0) {
3211
        if (!virNetworkObjIsActive(network)) {
3212
            virNetworkRemoveInactive(driver->networks, network);
3213 3214
            goto cleanup;
        }
3215 3216 3217 3218 3219
        /* if network was active already, just undo new persistent
         * definition by making it transient.
         * XXX - this isn't necessarily the correct thing to do.
         */
        virNetworkObjAssignDef(network, NULL, false);
3220 3221 3222
        goto cleanup;
    }

3223
    event = virNetworkEventLifecycleNew(def->name, def->uuid,
3224 3225
                                        VIR_NETWORK_EVENT_DEFINED,
                                        0);
3226

3227 3228
    VIR_INFO("Defining network '%s'", def->name);
    ret = virGetNetwork(conn, def->name, def->uuid);
3229

3230
 cleanup:
3231 3232
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3233
    if (freeDef)
J
Ján Tomko 已提交
3234
        virNetworkDefFree(def);
3235
    virNetworkObjEndAPI(&network);
3236
    return ret;
3237 3238
}

3239
static int
3240 3241
networkUndefine(virNetworkPtr net)
{
3242
    virNetworkDriverStatePtr driver = networkGetDriver();
3243
    virNetworkObjPtr network;
3244
    int ret = -1;
3245
    bool active = false;
3246
    virObjectEventPtr event = NULL;
3247

3248
    if (!(network = networkObjFromNetwork(net)))
3249
        goto cleanup;
3250

3251 3252 3253
    if (virNetworkUndefineEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3254 3255
    if (virNetworkObjIsActive(network))
        active = true;
3256

3257 3258 3259 3260 3261 3262
    if (!network->persistent) {
        virReportError(VIR_ERR_OPERATION_INVALID, "%s",
                       _("can't undefine transient network"));
        goto cleanup;
    }

3263
    /* remove autostart link */
3264
    if (virNetworkDeleteConfig(driver->networkConfigDir,
3265 3266
                               driver->networkAutostartDir,
                               network) < 0)
3267
        goto cleanup;
3268

3269 3270
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3271 3272
                                        VIR_NETWORK_EVENT_UNDEFINED,
                                        0);
3273

3274
    VIR_INFO("Undefining network '%s'", network->def->name);
3275
    if (!active) {
3276
        if (networkRemoveInactive(driver, network) < 0)
3277
            goto cleanup;
3278 3279 3280 3281 3282 3283
    } else {

        /* if the network still exists, it was active, and we need to make
         * it transient (by deleting the persistent def)
         */
        virNetworkObjAssignDef(network, NULL, false);
3284 3285
    }

3286
    ret = 0;
3287

3288
 cleanup:
3289 3290
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3291
    virNetworkObjEndAPI(&network);
3292
    return ret;
3293 3294
}

3295 3296 3297 3298 3299 3300 3301 3302
static int
networkUpdate(virNetworkPtr net,
              unsigned int command,
              unsigned int section,
              int parentIndex,
              const char *xml,
              unsigned int flags)
{
3303
    virNetworkDriverStatePtr driver = networkGetDriver();
3304
    virNetworkObjPtr network = NULL;
3305 3306
    int isActive, ret = -1;
    size_t i;
3307
    virNetworkIPDefPtr ipdef;
3308
    bool oldDhcpActive = false;
3309
    bool needFirewallRefresh = false;
3310

3311 3312 3313 3314 3315

    virCheckFlags(VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG,
                  -1);

3316
    if (!(network = networkObjFromNetwork(net)))
3317 3318
        goto cleanup;

3319 3320 3321
    if (virNetworkUpdateEnsureACL(net->conn, network->def, flags) < 0)
        goto cleanup;

3322
    /* see if we are listening for dhcp pre-modification */
3323
    for (i = 0;
3324
         (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET, i));
3325
         i++) {
3326 3327 3328 3329 3330 3331
        if (ipdef->nranges || ipdef->nhosts) {
            oldDhcpActive = true;
            break;
        }
    }

3332 3333
    /* VIR_NETWORK_UPDATE_AFFECT_CURRENT means "change LIVE if network
     * is active, else change CONFIG
J
Ján Tomko 已提交
3334
     */
3335
    isActive = virNetworkObjIsActive(network);
3336 3337
    if ((flags & (VIR_NETWORK_UPDATE_AFFECT_LIVE |
                  VIR_NETWORK_UPDATE_AFFECT_CONFIG)) ==
3338 3339 3340 3341 3342 3343 3344
        VIR_NETWORK_UPDATE_AFFECT_CURRENT) {
        if (isActive)
            flags |= VIR_NETWORK_UPDATE_AFFECT_LIVE;
        else
            flags |= VIR_NETWORK_UPDATE_AFFECT_CONFIG;
    }

3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361
    if (isActive && (flags & VIR_NETWORK_UPDATE_AFFECT_LIVE)) {
        /* Take care of anything that must be done before updating the
         * live NetworkDef.
         */
        if (network->def->forward.type == VIR_NETWORK_FORWARD_NONE ||
            network->def->forward.type == VIR_NETWORK_FORWARD_NAT ||
            network->def->forward.type == VIR_NETWORK_FORWARD_ROUTE) {
            switch (section) {
            case VIR_NETWORK_SECTION_FORWARD:
            case VIR_NETWORK_SECTION_FORWARD_INTERFACE:
            case VIR_NETWORK_SECTION_IP:
            case VIR_NETWORK_SECTION_IP_DHCP_RANGE:
            case VIR_NETWORK_SECTION_IP_DHCP_HOST:
                /* these could affect the firewall rules, so remove the
                 * old rules (and remember to load new ones after the
                 * update).
                 */
3362
                networkRemoveFirewallRules(network->def);
3363 3364 3365 3366 3367 3368 3369 3370
                needFirewallRefresh = true;
                break;
            default:
                break;
            }
        }
    }

3371
    /* update the network config in memory/on disk */
3372 3373
    if (virNetworkObjUpdate(network, command, section, parentIndex, xml, flags) < 0) {
        if (needFirewallRefresh)
3374
            ignore_value(networkAddFirewallRules(network->def));
3375 3376 3377
        goto cleanup;
    }

3378
    if (needFirewallRefresh && networkAddFirewallRules(network->def) < 0)
3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397
        goto cleanup;

    if (flags & VIR_NETWORK_UPDATE_AFFECT_CONFIG) {
        /* save updated persistent config to disk */
        if (virNetworkSaveConfig(driver->networkConfigDir,
                                 virNetworkObjGetPersistentDef(network)) < 0) {
            goto cleanup;
        }
    }

    if (isActive && (flags & VIR_NETWORK_UPDATE_AFFECT_LIVE)) {
        /* rewrite dnsmasq host files, restart dnsmasq, update iptables
         * rules, etc, according to which section was modified. Note that
         * some sections require multiple actions, so a single switch
         * statement is inadequate.
         */
        if (section == VIR_NETWORK_SECTION_BRIDGE ||
            section == VIR_NETWORK_SECTION_DOMAIN ||
            section == VIR_NETWORK_SECTION_IP ||
3398 3399 3400 3401 3402 3403 3404 3405
            section == VIR_NETWORK_SECTION_IP_DHCP_RANGE ||
            section == VIR_NETWORK_SECTION_DNS_TXT ||
            section == VIR_NETWORK_SECTION_DNS_SRV) {
            /* these sections all change things on the dnsmasq
             * commandline (i.e. in the .conf file), so we need to
             * kill and restart dnsmasq, because dnsmasq sets its uid
             * to "nobody" after it starts, and is unable to re-read
             * the conf file (owned by root, mode 600)
3406
             */
3407
            if (networkRestartDhcpDaemon(driver, network) < 0)
3408 3409
                goto cleanup;

3410 3411 3412 3413 3414 3415 3416 3417
        } else if (section == VIR_NETWORK_SECTION_IP_DHCP_HOST) {
            /* if we previously weren't listening for dhcp and now we
             * are (or vice-versa) then we need to do a restart,
             * otherwise we just need to do a refresh (redo the config
             * files and send SIGHUP)
             */
            bool newDhcpActive = false;

3418
            for (i = 0;
3419
                 (ipdef = virNetworkDefGetIPByIndex(network->def, AF_INET, i));
3420
                 i++) {
3421 3422 3423 3424 3425 3426 3427
                if (ipdef->nranges || ipdef->nhosts) {
                    newDhcpActive = true;
                    break;
                }
            }

            if ((newDhcpActive != oldDhcpActive &&
3428 3429
                 networkRestartDhcpDaemon(driver, network) < 0) ||
                networkRefreshDhcpDaemon(driver, network) < 0) {
3430 3431 3432
                goto cleanup;
            }

3433 3434 3435 3436
        } else if (section == VIR_NETWORK_SECTION_DNS_HOST) {
            /* this section only changes data in an external file
             * (not the .conf file) so we can just update the config
             * files and send SIGHUP to dnsmasq.
3437
             */
3438
            if (networkRefreshDhcpDaemon(driver, network) < 0)
3439 3440 3441 3442 3443 3444 3445 3446
                goto cleanup;

        }

        if (section == VIR_NETWORK_SECTION_IP) {
            /* only a change in IP addresses will affect radvd, and all of radvd's
             * config is stored in the conf file which will be re-read with a SIGHUP.
             */
3447
            if (networkRefreshRadvd(driver, network) < 0)
3448 3449 3450 3451
                goto cleanup;
        }

        /* save current network state to disk */
3452
        if ((ret = virNetworkSaveStatus(driver->stateDir,
3453
                                        network)) < 0) {
3454
            goto cleanup;
3455
        }
3456 3457
    }
    ret = 0;
3458
 cleanup:
3459
    virNetworkObjEndAPI(&network);
3460 3461 3462
    return ret;
}

3463 3464
static int networkCreate(virNetworkPtr net)
{
3465
    virNetworkDriverStatePtr driver = networkGetDriver();
3466 3467
    virNetworkObjPtr network;
    int ret = -1;
3468
    virObjectEventPtr event = NULL;
3469

3470
    if (!(network = networkObjFromNetwork(net)))
3471
        goto cleanup;
3472

3473 3474 3475
    if (virNetworkCreateEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3476
    if ((ret = networkStartNetwork(driver, network)) < 0)
3477
        goto cleanup;
3478

3479 3480
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3481 3482
                                        VIR_NETWORK_EVENT_STARTED,
                                        0);
3483

3484
 cleanup:
3485 3486
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3487
    virNetworkObjEndAPI(&network);
3488
    return ret;
3489 3490
}

3491 3492
static int networkDestroy(virNetworkPtr net)
{
3493
    virNetworkDriverStatePtr driver = networkGetDriver();
3494 3495
    virNetworkObjPtr network;
    int ret = -1;
3496
    virObjectEventPtr event = NULL;
3497

3498
    if (!(network = networkObjFromNetwork(net)))
3499
        goto cleanup;
3500

3501 3502 3503
    if (virNetworkDestroyEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

D
Daniel P. Berrange 已提交
3504
    if (!virNetworkObjIsActive(network)) {
3505
        virReportError(VIR_ERR_OPERATION_INVALID,
3506 3507
                       _("network '%s' is not active"),
                       network->def->name);
3508 3509 3510
        goto cleanup;
    }

3511
    if ((ret = networkShutdownNetwork(driver, network)) < 0)
3512 3513
        goto cleanup;

3514 3515
    event = virNetworkEventLifecycleNew(network->def->name,
                                        network->def->uuid,
3516 3517
                                        VIR_NETWORK_EVENT_STOPPED,
                                        0);
3518

3519
    if (!network->persistent &&
3520
        networkRemoveInactive(driver, network) < 0) {
3521 3522
        ret = -1;
        goto cleanup;
3523
    }
3524

3525
 cleanup:
3526 3527
    if (event)
        virObjectEventStateQueue(driver->networkEventState, event);
3528
    virNetworkObjEndAPI(&network);
3529 3530 3531
    return ret;
}

3532
static char *networkGetXMLDesc(virNetworkPtr net,
3533
                               unsigned int flags)
3534
{
3535
    virNetworkObjPtr network;
3536
    virNetworkDefPtr def;
3537
    char *ret = NULL;
3538

3539
    virCheckFlags(VIR_NETWORK_XML_INACTIVE, NULL);
3540

3541 3542
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3543

3544 3545 3546
    if (virNetworkGetXMLDescEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3547 3548 3549 3550 3551 3552
    if ((flags & VIR_NETWORK_XML_INACTIVE) && network->newDef)
        def = network->newDef;
    else
        def = network->def;

    ret = virNetworkDefFormat(def, flags);
3553

3554
 cleanup:
3555
    virNetworkObjEndAPI(&network);
3556
    return ret;
3557 3558 3559
}

static char *networkGetBridgeName(virNetworkPtr net) {
3560 3561 3562
    virNetworkObjPtr network;
    char *bridge = NULL;

3563 3564
    if (!(network = networkObjFromNetwork(net)))
        return bridge;
3565

3566 3567 3568
    if (virNetworkGetBridgeNameEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3569
    if (!(network->def->bridge)) {
3570 3571 3572
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("network '%s' does not have a bridge name."),
                       network->def->name);
3573 3574 3575
        goto cleanup;
    }

3576
    ignore_value(VIR_STRDUP(bridge, network->def->bridge));
3577

3578
 cleanup:
3579
    virNetworkObjEndAPI(&network);
3580 3581 3582 3583
    return bridge;
}

static int networkGetAutostart(virNetworkPtr net,
J
Ján Tomko 已提交
3584
                               int *autostart)
3585
{
3586 3587
    virNetworkObjPtr network;
    int ret = -1;
3588

3589 3590
    if (!(network = networkObjFromNetwork(net)))
        return ret;
3591

3592 3593 3594
    if (virNetworkGetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3595
    *autostart = network->autostart;
3596
    ret = 0;
3597

3598
 cleanup:
3599
    virNetworkObjEndAPI(&network);
3600
    return ret;
3601 3602 3603
}

static int networkSetAutostart(virNetworkPtr net,
3604 3605
                               int autostart)
{
3606
    virNetworkDriverStatePtr driver = networkGetDriver();
3607
    virNetworkObjPtr network;
3608
    char *configFile = NULL, *autostartLink = NULL;
3609
    int ret = -1;
3610

3611

3612
    if (!(network = networkObjFromNetwork(net)))
3613
        goto cleanup;
3614

3615 3616 3617
    if (virNetworkSetAutostartEnsureACL(net->conn, network->def) < 0)
        goto cleanup;

3618
    if (!network->persistent) {
3619 3620
        virReportError(VIR_ERR_OPERATION_INVALID,
                       "%s", _("cannot set autostart for transient network"));
3621 3622 3623
        goto cleanup;
    }

3624 3625
    autostart = (autostart != 0);

3626
    if (network->autostart != autostart) {
3627
        if ((configFile = virNetworkConfigFile(driver->networkConfigDir, network->def->name)) == NULL)
3628
            goto cleanup;
3629
        if ((autostartLink = virNetworkConfigFile(driver->networkAutostartDir, network->def->name)) == NULL)
3630 3631
            goto cleanup;

3632
        if (autostart) {
3633
            if (virFileMakePath(driver->networkAutostartDir) < 0) {
3634
                virReportSystemError(errno,
3635 3636
                                     _("cannot create autostart directory '%s'"),
                                     driver->networkAutostartDir);
3637 3638
                goto cleanup;
            }
3639

3640
            if (symlink(configFile, autostartLink) < 0) {
3641
                virReportSystemError(errno,
3642
                                     _("Failed to create symlink '%s' to '%s'"),
3643
                                     autostartLink, configFile);
3644 3645 3646
                goto cleanup;
            }
        } else {
3647
            if (unlink(autostartLink) < 0 && errno != ENOENT && errno != ENOTDIR) {
3648
                virReportSystemError(errno,
3649
                                     _("Failed to delete symlink '%s'"),
3650
                                     autostartLink);
3651 3652
                goto cleanup;
            }
3653 3654
        }

3655
        network->autostart = autostart;
3656
    }
3657
    ret = 0;
3658

3659
 cleanup:
3660 3661
    VIR_FREE(configFile);
    VIR_FREE(autostartLink);
3662
    virNetworkObjEndAPI(&network);
3663
    return ret;
3664 3665
}

3666
static int
3667 3668 3669 3670
networkGetDHCPLeases(virNetworkPtr network,
                     const char *mac,
                     virNetworkDHCPLeasePtr **leases,
                     unsigned int flags)
3671
{
3672
    virNetworkDriverStatePtr driver = networkGetDriver();
3673 3674 3675
    size_t i, j;
    size_t nleases = 0;
    int rv = -1;
3676
    ssize_t size = 0;
3677 3678 3679 3680 3681 3682 3683 3684 3685 3686 3687
    int custom_lease_file_len = 0;
    bool need_results = !!leases;
    long long currtime = 0;
    long long expirytime_tmp = -1;
    bool ipv6 = false;
    char *lease_entries = NULL;
    char *custom_lease_file = NULL;
    const char *ip_tmp = NULL;
    const char *mac_tmp = NULL;
    virJSONValuePtr lease_tmp = NULL;
    virJSONValuePtr leases_array = NULL;
3688
    virNetworkIPDefPtr ipdef_tmp = NULL;
3689 3690
    virNetworkDHCPLeasePtr lease = NULL;
    virNetworkDHCPLeasePtr *leases_ret = NULL;
3691
    virNetworkObjPtr obj;
3692
    virMacAddr mac_addr;
3693 3694 3695

    virCheckFlags(0, -1);

3696 3697 3698 3699 3700 3701
    /* only to check if the MAC is valid */
    if (mac && virMacAddrParse(mac, &mac_addr) < 0) {
        virReportError(VIR_ERR_INVALID_MAC, "%s", mac);
        return -1;
    }

3702 3703 3704 3705 3706
    if (!(obj = networkObjFromNetwork(network)))
        return -1;

    if (virNetworkGetDHCPLeasesEnsureACL(network->conn, obj->def) < 0)
        goto cleanup;
3707 3708

    /* Retrieve custom leases file location */
3709
    custom_lease_file = networkDnsmasqLeaseFileNameCustom(driver, obj->def->bridge);
3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752

    /* Read entire contents */
    if ((custom_lease_file_len = virFileReadAll(custom_lease_file,
                                                VIR_NETWORK_DHCP_LEASE_FILE_SIZE_MAX,
                                                &lease_entries)) < 0) {
        /* Even though src/network/leaseshelper.c guarantees the existence of
         * leases file (even if no leases are present), and the control reaches
         * here, instead of reporting error, return 0 leases */
        rv = 0;
        goto error;
    }

    if (custom_lease_file_len) {
        if (!(leases_array = virJSONValueFromString(lease_entries))) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("invalid json in file: %s"), custom_lease_file);
            goto error;
        }

        if ((size = virJSONValueArraySize(leases_array)) < 0) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("couldn't fetch array of leases"));
            goto error;
        }
    }

    currtime = (long long) time(NULL);

    for (i = 0; i < size; i++) {
        if (!(lease_tmp = virJSONValueArrayGet(leases_array, i))) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("failed to parse json"));
            goto error;
        }

        if (!(mac_tmp = virJSONValueObjectGetString(lease_tmp, "mac-address"))) {
            /* leaseshelper program guarantees that lease will be stored only if
             * mac-address is known otherwise not */
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("found lease without mac-address"));
            goto error;
        }

3753
        if (mac && virMacAddrCompare(mac, mac_tmp))
3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794
            continue;

        if (virJSONValueObjectGetNumberLong(lease_tmp, "expiry-time", &expirytime_tmp) < 0) {
            /* A lease cannot be present without expiry-time */
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("found lease without expiry-time"));
            goto error;
        }

        /* Do not report expired lease */
        if (expirytime_tmp < currtime)
            continue;

        if (need_results) {
            if (VIR_ALLOC(lease) < 0)
                goto error;

            lease->expirytime = expirytime_tmp;

            if (!(ip_tmp = virJSONValueObjectGetString(lease_tmp, "ip-address"))) {
                /* A lease without ip-address makes no sense */
                virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                               _("found lease without ip-address"));
                goto error;
            }

            /* Unlike IPv4, IPv6 uses ':' instead of '.' as separator */
            ipv6 = strchr(ip_tmp, ':') ? true : false;
            lease->type = ipv6 ? VIR_IP_ADDR_TYPE_IPV6 : VIR_IP_ADDR_TYPE_IPV4;

            /* Obtain prefix */
            for (j = 0; j < obj->def->nips; j++) {
                ipdef_tmp = &obj->def->ips[j];

                if (ipv6 && VIR_SOCKET_ADDR_IS_FAMILY(&ipdef_tmp->address,
                                                      AF_INET6)) {
                    lease->prefix = ipdef_tmp->prefix;
                    break;
                }
                if (!ipv6 && VIR_SOCKET_ADDR_IS_FAMILY(&ipdef_tmp->address,
                                                      AF_INET)) {
3795
                    lease->prefix = virSocketAddrGetIPPrefix(&ipdef_tmp->address,
3796 3797 3798 3799 3800 3801 3802 3803
                                                             &ipdef_tmp->netmask,
                                                             ipdef_tmp->prefix);
                    break;
                }
            }

            if ((VIR_STRDUP(lease->mac, mac_tmp) < 0) ||
                (VIR_STRDUP(lease->ipaddr, ip_tmp) < 0) ||
3804
                (VIR_STRDUP(lease->iface, obj->def->bridge) < 0))
3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836
                goto error;

            /* Fields that can be NULL */
            if ((VIR_STRDUP(lease->iaid,
                            virJSONValueObjectGetString(lease_tmp, "iaid")) < 0) ||
                (VIR_STRDUP(lease->clientid,
                            virJSONValueObjectGetString(lease_tmp, "client-id")) < 0) ||
                (VIR_STRDUP(lease->hostname,
                            virJSONValueObjectGetString(lease_tmp, "hostname")) < 0))
                goto error;

            if (VIR_INSERT_ELEMENT(leases_ret, nleases, nleases, lease) < 0)
                goto error;

        } else {
            nleases++;
        }

        VIR_FREE(lease);
    }

    if (leases_ret) {
        /* NULL terminated array */
        ignore_value(VIR_REALLOC_N(leases_ret, nleases + 1));
        *leases = leases_ret;
        leases_ret = NULL;
    }

    rv = nleases;

 cleanup:
    VIR_FREE(lease);
3837
    VIR_FREE(lease_entries);
3838 3839
    VIR_FREE(custom_lease_file);
    virJSONValueFree(leases_array);
3840

3841
    virNetworkObjEndAPI(&obj);
3842

3843 3844 3845 3846 3847 3848 3849 3850 3851 3852 3853
    return rv;

 error:
    if (leases_ret) {
        for (i = 0; i < nleases; i++)
            virNetworkDHCPLeaseFree(leases_ret[i]);
        VIR_FREE(leases_ret);
    }
    goto cleanup;
}

3854 3855

static virNetworkDriver networkDriver = {
3856
    .name = "bridge",
3857 3858 3859 3860 3861
    .connectNumOfNetworks = networkConnectNumOfNetworks, /* 0.2.0 */
    .connectListNetworks = networkConnectListNetworks, /* 0.2.0 */
    .connectNumOfDefinedNetworks = networkConnectNumOfDefinedNetworks, /* 0.2.0 */
    .connectListDefinedNetworks = networkConnectListDefinedNetworks, /* 0.2.0 */
    .connectListAllNetworks = networkConnectListAllNetworks, /* 0.10.2 */
3862 3863
    .connectNetworkEventRegisterAny = networkConnectNetworkEventRegisterAny, /* 1.2.1 */
    .connectNetworkEventDeregisterAny = networkConnectNetworkEventDeregisterAny, /* 1.2.1 */
3864 3865
    .networkLookupByUUID = networkLookupByUUID, /* 0.2.0 */
    .networkLookupByName = networkLookupByName, /* 0.2.0 */
3866 3867
    .networkCreateXML = networkCreateXML, /* 0.2.0 */
    .networkDefineXML = networkDefineXML, /* 0.2.0 */
3868
    .networkUndefine = networkUndefine, /* 0.2.0 */
3869
    .networkUpdate = networkUpdate, /* 0.10.2 */
3870
    .networkCreate = networkCreate, /* 0.2.0 */
3871 3872 3873 3874 3875 3876 3877
    .networkDestroy = networkDestroy, /* 0.2.0 */
    .networkGetXMLDesc = networkGetXMLDesc, /* 0.2.0 */
    .networkGetBridgeName = networkGetBridgeName, /* 0.2.0 */
    .networkGetAutostart = networkGetAutostart, /* 0.2.1 */
    .networkSetAutostart = networkSetAutostart, /* 0.2.1 */
    .networkIsActive = networkIsActive, /* 0.7.3 */
    .networkIsPersistent = networkIsPersistent, /* 0.7.3 */
3878
    .networkGetDHCPLeases = networkGetDHCPLeases, /* 1.2.6 */
3879 3880 3881
};

static virStateDriver networkStateDriver = {
3882
    .name = "bridge",
3883
    .stateInitialize  = networkStateInitialize,
3884
    .stateAutoStart  = networkStateAutoStart,
3885 3886
    .stateCleanup = networkStateCleanup,
    .stateReload = networkStateReload,
3887 3888
};

3889 3890
int networkRegister(void)
{
3891
    if (virSetSharedNetworkDriver(&networkDriver) < 0)
3892
        return -1;
3893 3894
    if (virRegisterStateDriver(&networkStateDriver) < 0)
        return -1;
3895 3896
    return 0;
}
3897 3898 3899

/********************************************************/

3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934
/* A unified function to log network connections and disconnections */

static void
networkLogAllocation(virNetworkDefPtr netdef,
                     virDomainNetType actualType,
                     virNetworkForwardIfDefPtr dev,
                     virDomainNetDefPtr iface,
                     bool inUse)
{
    char macStr[VIR_MAC_STRING_BUFLEN];
    const char *verb = inUse ? "using" : "releasing";

    if (!dev) {
        VIR_INFO("MAC %s %s network %s (%d connections)",
                 virMacAddrFormat(&iface->mac, macStr), verb,
                 netdef->name, netdef->connections);
    } else {
        if (actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV) {
            VIR_INFO("MAC %s %s network %s (%d connections) "
                     "physical device %04x:%02x:%02x.%x (%d connections)",
                     virMacAddrFormat(&iface->mac, macStr), verb,
                     netdef->name, netdef->connections,
                     dev->device.pci.domain, dev->device.pci.bus,
                     dev->device.pci.slot, dev->device.pci.function,
                     dev->connections);
        } else {
            VIR_INFO("MAC %s %s network %s (%d connections) "
                     "physical device %s (%d connections)",
                     virMacAddrFormat(&iface->mac, macStr), verb,
                     netdef->name, netdef->connections,
                     dev->device.dev, dev->connections);
        }
    }
}

3935 3936 3937 3938 3939 3940 3941 3942 3943
/* Private API to deal with logical switch capabilities.
 * These functions are exported so that other parts of libvirt can
 * call them, but are not part of the public API and not in the
 * driver's function table. If we ever have more than one network
 * driver, we will need to present these functions via a second
 * "backend" function table.
 */

/* networkAllocateActualDevice:
3944
 * @dom: domain definition that @iface belongs to
3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955
 * @iface: the original NetDef from the domain
 *
 * Looks up the network reference by iface, allocates a physical
 * device from that network (if appropriate), and returns with the
 * virDomainActualNetDef filled in accordingly. If there are no
 * changes to be made in the netdef, then just leave the actualdef
 * empty.
 *
 * Returns 0 on success, -1 on failure.
 */
int
3956 3957
networkAllocateActualDevice(virDomainDefPtr dom,
                            virDomainNetDefPtr iface)
3958
{
3959
    virNetworkDriverStatePtr driver = networkGetDriver();
3960
    virDomainNetType actualType = iface->type;
3961 3962
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
3963
    virNetDevBandwidthPtr bandwidth = NULL;
3964 3965 3966
    virPortGroupDefPtr portgroup = NULL;
    virNetDevVPortProfilePtr virtport = iface->virtPortProfile;
    virNetDevVlanPtr vlan = NULL;
3967
    virNetworkForwardIfDefPtr dev = NULL;
3968
    size_t i;
3969 3970 3971
    int ret = -1;

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
3972
        goto validate;
3973 3974 3975 3976

    virDomainActualNetDefFree(iface->data.network.actual);
    iface->data.network.actual = NULL;

3977
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
3978
    if (!network) {
3979 3980 3981
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
3982
        goto error;
3983 3984
    }
    netdef = network->def;
3985

3986 3987 3988 3989 3990 3991 3992
    if (!virNetworkObjIsActive(network)) {
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("network '%s' is not active"),
                       netdef->name);
        goto error;
    }

3993 3994 3995
    if (VIR_ALLOC(iface->data.network.actual) < 0)
        goto error;

3996 3997 3998
    /* portgroup can be present for any type of network, in particular
     * for bandwidth information, so we need to check for that and
     * fill it in appropriately for all forward types.
J
Ján Tomko 已提交
3999
     */
4000 4001 4002 4003 4004 4005
    portgroup = virPortGroupFindByName(netdef, iface->data.network.portgroup);

    /* If there is already interface-specific bandwidth, just use that
     * (already in NetDef). Otherwise, if there is bandwidth info in
     * the portgroup, fill that into the ActualDef.
     */
4006 4007 4008 4009 4010 4011

    if (iface->bandwidth)
        bandwidth = iface->bandwidth;
    else if (portgroup && portgroup->bandwidth)
        bandwidth = portgroup->bandwidth;

4012 4013
    if (bandwidth && virNetDevBandwidthCopy(&iface->data.network.actual->bandwidth,
                                            bandwidth) < 0)
4014
        goto error;
4015

4016 4017 4018 4019 4020 4021 4022 4023
    /* copy appropriate vlan info to actualNet */
    if (iface->vlan.nTags > 0)
        vlan = &iface->vlan;
    else if (portgroup && portgroup->vlan.nTags > 0)
        vlan = &portgroup->vlan;
    else if (netdef->vlan.nTags > 0)
        vlan = &netdef->vlan;

4024 4025
    if (vlan && virNetDevVlanCopy(&iface->data.network.actual->vlan, vlan) < 0)
        goto error;
4026

4027 4028 4029 4030 4031 4032 4033 4034 4035 4036
    if (iface->trustGuestRxFilters)
       iface->data.network.actual->trustGuestRxFilters
          = iface->trustGuestRxFilters;
    else if (portgroup && portgroup->trustGuestRxFilters)
       iface->data.network.actual->trustGuestRxFilters
          = portgroup->trustGuestRxFilters;
    else if (netdef->trustGuestRxFilters)
       iface->data.network.actual->trustGuestRxFilters
          = netdef->trustGuestRxFilters;

4037 4038 4039
    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE)) {
4040
        /* for these forward types, the actual net type really *is*
4041
         * NETWORK; we just keep the info from the portgroup in
4042
         * iface->data.network.actual
J
Ján Tomko 已提交
4043
         */
4044
        iface->data.network.actual->type = VIR_DOMAIN_NET_TYPE_NETWORK;
4045

4046
        /* we also store the bridge device and macTableManager settings
4047 4048 4049 4050 4051 4052 4053
         * in iface->data.network.actual->data.bridge for later use
         * after the domain's tap device is created (to attach to the
         * bridge and set flood/learning mode on the tap device)
         */
        if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                       netdef->bridge) < 0)
            goto error;
4054 4055
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
4056

4057 4058 4059
        if (networkPlugBandwidth(network, iface) < 0)
            goto error;

4060
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
4061
               netdef->bridge) {
4062 4063 4064 4065 4066

        /* <forward type='bridge'/> <bridge name='xxx'/>
         * is VIR_DOMAIN_NET_TYPE_BRIDGE
         */

4067
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_BRIDGE;
4068 4069
        if (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                       netdef->bridge) < 0)
4070
            goto error;
4071 4072
        iface->data.network.actual->data.bridge.macTableManager
           = netdef->macTableManager;
4073

4074 4075 4076 4077 4078 4079 4080 4081
        /* merge virtualports from interface, network, and portgroup to
         * arrive at actual virtualport to use
         */
        if (virNetDevVPortProfileMerge3(&iface->data.network.actual->virtPortProfile,
                                        iface->virtPortProfile,
                                        netdef->virtPortProfile,
                                        portgroup
                                        ? portgroup->virtPortProfile : NULL) < 0) {
4082
            goto error;
4083 4084 4085 4086 4087 4088 4089 4090 4091 4092
        }
        virtport = iface->data.network.actual->virtPortProfile;
        if (virtport) {
            /* only type='openvswitch' is allowed for bridges */
            if (virtport->virtPortType != VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("<virtualport type='%s'> not supported for network "
                                 "'%s' which uses a bridge device"),
                               virNetDevVPortTypeToString(virtport->virtPortType),
                               netdef->name);
4093
                goto error;
4094 4095 4096
            }
        }

4097
    } else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4098

4099
        virDomainHostdevSubsysPCIBackendType backend;
4100

4101
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_HOSTDEV;
4102
        if (networkCreateInterfacePool(netdef) < 0)
4103 4104 4105
            goto error;

        /* pick first dev with 0 connections */
4106 4107 4108
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].connections == 0) {
                dev = &netdef->forward.ifs[i];
4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122
                break;
            }
        }
        if (!dev) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' requires exclusive access "
                             "to interfaces, but none are available"),
                           netdef->name);
            goto error;
        }
        iface->data.network.actual->data.hostdev.def.parent.type = VIR_DOMAIN_DEVICE_NET;
        iface->data.network.actual->data.hostdev.def.parent.data.net = iface;
        iface->data.network.actual->data.hostdev.def.info = &iface->info;
        iface->data.network.actual->data.hostdev.def.mode = VIR_DOMAIN_HOSTDEV_MODE_SUBSYS;
4123
        iface->data.network.actual->data.hostdev.def.managed = netdef->forward.managed ? 1 : 0;
4124
        iface->data.network.actual->data.hostdev.def.source.subsys.type = dev->type;
4125
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.addr = dev->device.pci;
4126

E
Eric Blake 已提交
4127
        switch (netdef->forward.driverName) {
4128
        case VIR_NETWORK_FORWARD_DRIVER_NAME_DEFAULT:
4129
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_DEFAULT;
4130 4131
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_KVM:
4132
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_KVM;
4133 4134
            break;
        case VIR_NETWORK_FORWARD_DRIVER_NAME_VFIO:
4135
            backend = VIR_DOMAIN_HOSTDEV_PCI_BACKEND_VFIO;
4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146
            break;
        default:
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("unrecognized driver name value %d "
                             " in network '%s'"),
                           netdef->forward.driverName, netdef->name);
            goto error;
        }
        iface->data.network.actual->data.hostdev.def.source.subsys.u.pci.backend
            = backend;

4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171
        /* merge virtualports from interface, network, and portgroup to
         * arrive at actual virtualport to use
         */
        if (virNetDevVPortProfileMerge3(&iface->data.network.actual->virtPortProfile,
                                        iface->virtPortProfile,
                                        netdef->virtPortProfile,
                                        portgroup
                                        ? portgroup->virtPortProfile : NULL) < 0) {
            goto error;
        }
        virtport = iface->data.network.actual->virtPortProfile;
        if (virtport) {
            /* make sure type is supported for hostdev connections */
            if (virtport->virtPortType != VIR_NETDEV_VPORT_PROFILE_8021QBG &&
                virtport->virtPortType != VIR_NETDEV_VPORT_PROFILE_8021QBH) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("<virtualport type='%s'> not supported for network "
                                 "'%s' which uses an SR-IOV Virtual Function "
                                 "via PCI passthrough"),
                               virNetDevVPortTypeToString(virtport->virtPortType),
                               netdef->name);
                goto error;
            }
        }

4172 4173 4174 4175
    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) ||
               (netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) ||
               (netdef->forward.type == VIR_NETWORK_FORWARD_VEPA) ||
               (netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH)) {
4176 4177 4178 4179 4180 4181

        /* <forward type='bridge|private|vepa|passthrough'> are all
         * VIR_DOMAIN_NET_TYPE_DIRECT.
         */

        /* Set type=direct and appropriate <source mode='xxx'/> */
4182
        iface->data.network.actual->type = actualType = VIR_DOMAIN_NET_TYPE_DIRECT;
4183
        switch (netdef->forward.type) {
4184
        case VIR_NETWORK_FORWARD_BRIDGE:
4185
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_BRIDGE;
4186 4187
            break;
        case VIR_NETWORK_FORWARD_PRIVATE:
4188
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PRIVATE;
4189 4190
            break;
        case VIR_NETWORK_FORWARD_VEPA:
4191
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_VEPA;
4192 4193
            break;
        case VIR_NETWORK_FORWARD_PASSTHROUGH:
4194
            iface->data.network.actual->data.direct.mode = VIR_NETDEV_MACVLAN_MODE_PASSTHRU;
4195 4196 4197
            break;
        }

4198 4199 4200 4201 4202 4203 4204 4205
        /* merge virtualports from interface, network, and portgroup to
         * arrive at actual virtualport to use
         */
        if (virNetDevVPortProfileMerge3(&iface->data.network.actual->virtPortProfile,
                                        iface->virtPortProfile,
                                        netdef->virtPortProfile,
                                        portgroup
                                        ? portgroup->virtPortProfile : NULL) < 0) {
4206
            goto error;
4207
        }
4208
        virtport = iface->data.network.actual->virtPortProfile;
4209
        if (virtport) {
4210 4211 4212 4213 4214 4215 4216 4217
            /* make sure type is supported for macvtap connections */
            if (virtport->virtPortType != VIR_NETDEV_VPORT_PROFILE_8021QBG &&
                virtport->virtPortType != VIR_NETDEV_VPORT_PROFILE_8021QBH) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("<virtualport type='%s'> not supported for network "
                                 "'%s' which uses a macvtap device"),
                               virNetDevVPortTypeToString(virtport->virtPortType),
                               netdef->name);
4218
                goto error;
4219 4220
            }
        }
4221

4222 4223 4224
        /* If there is only a single device, just return it (caller will detect
         * any error if exclusive use is required but could not be acquired).
         */
4225
        if ((netdef->forward.nifs <= 0) && (netdef->forward.npfs <= 0)) {
4226 4227 4228 4229
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' uses a direct mode, but "
                             "has no forward dev and no interface pool"),
                           netdef->name);
4230
            goto error;
4231 4232 4233
        } else {
            /* pick an interface from the pool */

4234
            if (networkCreateInterfacePool(netdef) < 0)
4235 4236
                goto error;

4237 4238 4239 4240 4241
            /* PASSTHROUGH mode, and PRIVATE Mode + 802.1Qbh both
             * require exclusive access to a device, so current
             * connections count must be 0.  Other modes can share, so
             * just search for the one with the lowest number of
             * connections.
4242
             */
4243 4244
            if ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
                ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4245 4246 4247
                 iface->data.network.actual->virtPortProfile &&
                 (iface->data.network.actual->virtPortProfile->virtPortType
                  == VIR_NETDEV_VPORT_PROFILE_8021QBH))) {
4248

4249
                /* pick first dev with 0 connections */
4250 4251 4252
                for (i = 0; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections == 0) {
                        dev = &netdef->forward.ifs[i];
4253 4254 4255 4256 4257
                        break;
                    }
                }
            } else {
                /* pick least used dev */
4258
                dev = &netdef->forward.ifs[0];
4259 4260 4261
                for (i = 1; i < netdef->forward.nifs; i++) {
                    if (netdef->forward.ifs[i].connections < dev->connections)
                        dev = &netdef->forward.ifs[i];
4262 4263 4264 4265
                }
            }
            /* dev points at the physical device we want to use */
            if (!dev) {
4266 4267 4268 4269
                virReportError(VIR_ERR_INTERNAL_ERROR,
                               _("network '%s' requires exclusive access "
                                 "to interfaces, but none are available"),
                               netdef->name);
4270
                goto error;
4271
            }
4272 4273
            if (VIR_STRDUP(iface->data.network.actual->data.direct.linkdev,
                           dev->device.dev) < 0)
4274
                goto error;
4275 4276 4277
        }
    }

4278
    if (virNetDevVPortProfileCheckComplete(virtport, true) < 0)
4279
        goto error;
4280

4281
 validate:
4282 4283 4284 4285 4286
    /* make sure that everything now specified for the device is
     * actually supported on this type of network. NB: network,
     * netdev, and iface->data.network.actual may all be NULL.
     */

4287
    if (virDomainNetGetActualVlan(iface)) {
4288 4289 4290 4291
        /* vlan configuration via libvirt is only supported for PCI
         * Passthrough SR-IOV devices (hostdev or macvtap passthru
         * mode) and openvswitch bridges. Otherwise log an error and
         * fail
4292 4293
         */
        if (!(actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV ||
4294 4295 4296
              (actualType == VIR_DOMAIN_NET_TYPE_DIRECT &&
               virDomainNetGetActualDirectMode(iface)
               == VIR_NETDEV_MACVLAN_MODE_PASSTHRU) ||
4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310 4311 4312 4313 4314 4315
              (actualType == VIR_DOMAIN_NET_TYPE_BRIDGE &&
               virtport && virtport->virtPortType
               == VIR_NETDEV_VPORT_PROFILE_OPENVSWITCH))) {
            if (netdef) {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("an interface connecting to network '%s' "
                                 "is requesting a vlan tag, but that is not "
                                 "supported for this type of network"),
                               netdef->name);
            } else {
                virReportError(VIR_ERR_CONFIG_UNSUPPORTED,
                               _("an interface of type '%s' "
                                 "is requesting a vlan tag, but that is not "
                                 "supported for this type of connection"),
                               virDomainNetTypeToString(iface->type));
            }
            goto error;
        }
    }
4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326
    if (virDomainNetGetActualBandwidth(iface)) {
        /* bandwidth configuration via libvirt is not supported for
         * hostdev network devices
         */
        if (actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV) {
            virReportError(VIR_ERR_CONFIG_UNSUPPORTED, "%s",
                           _("bandwidth settings are not supported "
                             "for hostdev interfaces"));
            goto error;
        }
    }
4327 4328 4329

    if (netdef) {
        netdef->connections++;
4330
        if (dev)
4331 4332 4333 4334 4335 4336
            dev->connections++;
        /* finally we can call the 'plugged' hook script if any */
        if (networkRunHook(network, dom, iface,
                           VIR_HOOK_NETWORK_OP_IFACE_PLUGGED,
                           VIR_HOOK_SUBOP_BEGIN) < 0) {
            /* adjust for failure */
4337
            netdef->connections--;
4338 4339 4340 4341
            if (dev)
                dev->connections--;
            goto error;
        }
4342
        networkLogAllocation(netdef, actualType, dev, iface, true);
4343 4344
    }

4345
    ret = 0;
4346

4347
 cleanup:
4348
    virNetworkObjEndAPI(&network);
4349 4350
    return ret;

4351
 error:
4352
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
4353 4354 4355
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4356
    goto cleanup;
4357 4358 4359
}

/* networkNotifyActualDevice:
4360
 * @dom: domain definition that @iface belongs to
4361 4362 4363 4364 4365 4366 4367 4368 4369 4370
 * @iface:  the domain's NetDef with an "actual" device already filled in.
 *
 * Called to notify the network driver when libvirtd is restarted and
 * finds an already running domain. If appropriate it will force an
 * allocation of the actual->direct.linkdev to get everything back in
 * order.
 *
 * Returns 0 on success, -1 on failure.
 */
int
4371 4372
networkNotifyActualDevice(virDomainDefPtr dom,
                          virDomainNetDefPtr iface)
4373
{
4374
    virNetworkDriverStatePtr driver = networkGetDriver();
4375
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4376 4377
    virNetworkObjPtr network;
    virNetworkDefPtr netdef;
4378
    virNetworkForwardIfDefPtr dev = NULL;
4379 4380
    size_t i;
    int ret = -1;
4381 4382

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4383
        return 0;
4384

4385
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4386
    if (!network) {
4387 4388 4389
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4390 4391 4392 4393
        goto error;
    }
    netdef = network->def;

4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404
    /* if we're restarting libvirtd after an upgrade from a version
     * that didn't save bridge name in actualNetDef for
     * actualType==network, we need to copy it in so that it will be
     * available in all cases
     */
    if (actualType == VIR_DOMAIN_NET_TYPE_NETWORK &&
        !iface->data.network.actual->data.bridge.brname &&
        (VIR_STRDUP(iface->data.network.actual->data.bridge.brname,
                    netdef->bridge) < 0))
            goto error;

4405
    if (!iface->data.network.actual ||
4406 4407
        (actualType != VIR_DOMAIN_NET_TYPE_DIRECT &&
         actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV)) {
4408 4409
        VIR_DEBUG("Nothing to claim from network %s", iface->data.network.name);
        goto success;
4410 4411
    }

4412
    if (networkCreateInterfacePool(netdef) < 0)
4413
        goto error;
4414

4415
    if (netdef->forward.nifs == 0) {
4416
        virReportError(VIR_ERR_INTERNAL_ERROR,
4417 4418
                       _("network '%s' uses a direct or hostdev mode, "
                         "but has no forward dev and no interface pool"),
4419
                       netdef->name);
4420
        goto error;
4421
    }
4422

4423 4424
    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;
4425

4426 4427 4428 4429 4430 4431 4432 4433 4434
        actualDev = virDomainNetGetActualDirectDev(iface);
        if (!actualDev) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("the interface uses a direct mode, "
                             "but has no source dev"));
            goto error;
        }

        /* find the matching interface and increment its connections */
4435 4436
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4437
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4438 4439
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4440 4441 4442 4443 4444
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
4445
            virReportError(VIR_ERR_INTERNAL_ERROR,
4446 4447
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4448
                           netdef->name, actualDev);
4449
            goto error;
4450 4451
        }

4452
        /* PASSTHROUGH mode and PRIVATE Mode + 802.1Qbh both require
4453 4454
         * exclusive access to a device, so current connections count
         * must be 0 in those cases.
4455
         */
4456
        if ((dev->connections > 0) &&
4457 4458
            ((netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH) ||
             ((netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) &&
4459 4460
              iface->data.network.actual->virtPortProfile &&
              (iface->data.network.actual->virtPortProfile->virtPortType
4461
               == VIR_NETDEV_VPORT_PROFILE_8021QBH)))) {
4462
            virReportError(VIR_ERR_INTERNAL_ERROR,
4463 4464
                           _("network '%s' claims dev='%s' is already in "
                             "use by a different domain"),
4465
                           netdef->name, actualDev);
4466
            goto error;
4467
        }
4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479
    }  else /* if (actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV) */ {
        virDomainHostdevDefPtr hostdev;

        hostdev = virDomainNetGetActualHostdev(iface);
        if (!hostdev) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("the interface uses a hostdev mode, "
                             "but has no hostdev"));
            goto error;
        }

        /* find the matching interface and increment its connections */
4480 4481
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4482
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4483
                virPCIDeviceAddressEqual(&hostdev->source.subsys.u.pci.addr,
4484 4485
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4486 4487 4488 4489 4490 4491 4492 4493 4494
                break;
            }
        }
        /* dev points at the physical device we want to use */
        if (!dev) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' doesn't have "
                             "PCI device %04x:%02x:%02x.%x in use by domain"),
                           netdef->name,
4495 4496 4497 4498
                           hostdev->source.subsys.u.pci.addr.domain,
                           hostdev->source.subsys.u.pci.addr.bus,
                           hostdev->source.subsys.u.pci.addr.slot,
                           hostdev->source.subsys.u.pci.addr.function);
J
Ján Tomko 已提交
4499
            goto error;
4500 4501 4502 4503 4504 4505 4506
        }

        /* PASSTHROUGH mode, PRIVATE Mode + 802.1Qbh, and hostdev (PCI
         * passthrough) all require exclusive access to a device, so
         * current connections count must be 0 in those cases.
         */
        if ((dev->connections > 0) &&
4507
            netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {
4508 4509 4510 4511 4512 4513 4514 4515 4516
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' claims the PCI device at "
                             "domain=%d bus=%d slot=%d function=%d "
                             "is already in use by a different domain"),
                           netdef->name,
                           dev->device.pci.domain, dev->device.pci.bus,
                           dev->device.pci.slot, dev->device.pci.function);
            goto error;
        }
4517 4518
    }

4519
 success:
4520
    netdef->connections++;
4521 4522
    if (dev)
        dev->connections++;
4523 4524 4525 4526 4527 4528 4529 4530 4531
    /* finally we can call the 'plugged' hook script if any */
    if (networkRunHook(network, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_PLUGGED,
                       VIR_HOOK_SUBOP_BEGIN) < 0) {
        /* adjust for failure */
        if (dev)
            dev->connections--;
        netdef->connections--;
        goto error;
    }
4532
    networkLogAllocation(netdef, actualType, dev, iface, true);
4533

4534
    ret = 0;
4535
 cleanup:
4536
    virNetworkObjEndAPI(&network);
4537
    return ret;
4538

4539
 error:
4540
    goto cleanup;
4541 4542 4543
}


4544

4545
/* networkReleaseActualDevice:
4546
 * @dom: domain definition that @iface belongs to
4547 4548 4549 4550 4551 4552 4553 4554 4555 4556
 * @iface:  a domain's NetDef (interface definition)
 *
 * Given a domain <interface> element that previously had its <actual>
 * element filled in (and possibly a physical device allocated to it),
 * free up the physical device for use by someone else, and free the
 * virDomainActualNetDef.
 *
 * Returns 0 on success, -1 on failure.
 */
int
4557 4558
networkReleaseActualDevice(virDomainDefPtr dom,
                           virDomainNetDefPtr iface)
4559
{
4560
    virNetworkDriverStatePtr driver = networkGetDriver();
4561
    virDomainNetType actualType = virDomainNetGetActualType(iface);
4562
    virNetworkObjPtr network;
4563
    virNetworkDefPtr netdef;
4564
    virNetworkForwardIfDefPtr dev = NULL;
4565 4566
    size_t i;
    int ret = -1;
4567 4568

    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
J
Ján Tomko 已提交
4569
        return 0;
4570

4571
    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
4572
    if (!network) {
4573 4574 4575
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
4576 4577 4578 4579
        goto error;
    }
    netdef = network->def;

4580 4581
    if (iface->data.network.actual &&
        (netdef->forward.type == VIR_NETWORK_FORWARD_NONE ||
4582 4583 4584 4585 4586
         netdef->forward.type == VIR_NETWORK_FORWARD_NAT ||
         netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE) &&
        networkUnplugBandwidth(network, iface) < 0)
        goto error;

4587 4588 4589
    if ((!iface->data.network.actual) ||
        ((actualType != VIR_DOMAIN_NET_TYPE_DIRECT) &&
         (actualType != VIR_DOMAIN_NET_TYPE_HOSTDEV))) {
4590 4591
        VIR_DEBUG("Nothing to release to network %s", iface->data.network.name);
        goto success;
4592 4593
    }

4594
    if (netdef->forward.nifs == 0) {
4595
        virReportError(VIR_ERR_INTERNAL_ERROR,
4596
                       _("network '%s' uses a direct/hostdev mode, but "
4597 4598
                         "has no forward dev and no interface pool"),
                       netdef->name);
4599
        goto error;
4600 4601 4602 4603 4604 4605 4606 4607 4608 4609 4610 4611
    }

    if (actualType == VIR_DOMAIN_NET_TYPE_DIRECT) {
        const char *actualDev;

        actualDev = virDomainNetGetActualDirectDev(iface);
        if (!actualDev) {
            virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                           _("the interface uses a direct mode, "
                             "but has no source dev"));
            goto error;
        }
4612

4613 4614
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4615
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_NETDEV &&
4616 4617
                STREQ(actualDev, netdef->forward.ifs[i].device.dev)) {
                dev = &netdef->forward.ifs[i];
4618 4619 4620
                break;
            }
        }
4621

4622
        if (!dev) {
4623
            virReportError(VIR_ERR_INTERNAL_ERROR,
4624 4625
                           _("network '%s' doesn't have dev='%s' "
                             "in use by domain"),
4626
                           netdef->name, actualDev);
4627
            goto error;
4628
        }
4629 4630 4631 4632 4633 4634 4635 4636 4637 4638
    } else /* if (actualType == VIR_DOMAIN_NET_TYPE_HOSTDEV) */ {
        virDomainHostdevDefPtr hostdev;

        hostdev = virDomainNetGetActualHostdev(iface);
        if (!hostdev) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           "%s", _("the interface uses a hostdev mode, but has no hostdev"));
            goto error;
        }

4639 4640
        for (i = 0; i < netdef->forward.nifs; i++) {
            if (netdef->forward.ifs[i].type
4641
                == VIR_NETWORK_FORWARD_HOSTDEV_DEVICE_PCI &&
4642
                virPCIDeviceAddressEqual(&hostdev->source.subsys.u.pci.addr,
4643 4644
                                         &netdef->forward.ifs[i].device.pci)) {
                dev = &netdef->forward.ifs[i];
4645 4646 4647 4648 4649 4650 4651 4652 4653
                break;
            }
        }

        if (!dev) {
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' doesn't have "
                             "PCI device %04x:%02x:%02x.%x in use by domain"),
                           netdef->name,
4654 4655 4656 4657
                           hostdev->source.subsys.u.pci.addr.domain,
                           hostdev->source.subsys.u.pci.addr.bus,
                           hostdev->source.subsys.u.pci.addr.slot,
                           hostdev->source.subsys.u.pci.addr.function);
J
Ján Tomko 已提交
4658
            goto error;
4659
        }
J
Ján Tomko 已提交
4660
    }
4661

4662
 success:
4663
    if (iface->data.network.actual) {
4664
        netdef->connections--;
4665 4666
        if (dev)
            dev->connections--;
4667 4668 4669
        /* finally we can call the 'unplugged' hook script if any */
        networkRunHook(network, dom, iface, VIR_HOOK_NETWORK_OP_IFACE_UNPLUGGED,
                       VIR_HOOK_SUBOP_BEGIN);
4670
        networkLogAllocation(netdef, actualType, dev, iface, false);
4671
    }
4672
    ret = 0;
4673
 cleanup:
4674
    virNetworkObjEndAPI(&network);
4675 4676 4677 4678
    if (iface->type == VIR_DOMAIN_NET_TYPE_NETWORK) {
        virDomainActualNetDefFree(iface->data.network.actual);
        iface->data.network.actual = NULL;
    }
4679
    return ret;
4680

4681
 error:
4682
    goto cleanup;
4683
}
4684 4685 4686 4687 4688 4689

/*
 * networkGetNetworkAddress:
 * @netname: the name of a network
 * @netaddr: string representation of IP address for that network.
 *
4690
 * Attempt to return an IP address associated with the named
4691 4692 4693 4694
 * network. If a libvirt virtual network, that will be provided in the
 * configuration. For host bridge and direct (macvtap) networks, we
 * must do an ioctl to learn the address.
 *
4695
 * Note: This function returns the first IP address it finds. It might
4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706
 * be useful if it was more flexible, but the current use (getting a
 * listen address for qemu's vnc/spice graphics server) can only use a
 * single address anyway.
 *
 * Returns 0 on success, and puts a string (which must be free'd by
 * the caller) into *netaddr. Returns -1 on failure or -2 if
 * completely unsupported.
 */
int
networkGetNetworkAddress(const char *netname, char **netaddr)
{
4707
    virNetworkDriverStatePtr driver = networkGetDriver();
4708
    int ret = -1;
4709
    virNetworkObjPtr network;
4710
    virNetworkDefPtr netdef;
4711
    virNetworkIPDefPtr ipdef;
4712 4713
    virSocketAddr addr;
    virSocketAddrPtr addrptr = NULL;
4714
    char *dev_name = NULL;
4715 4716

    *netaddr = NULL;
4717
    network = virNetworkObjFindByName(driver->networks, netname);
4718
    if (!network) {
4719 4720 4721
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       netname);
4722
        goto cleanup;
4723 4724 4725
    }
    netdef = network->def;

4726
    switch (netdef->forward.type) {
4727 4728 4729
    case VIR_NETWORK_FORWARD_NONE:
    case VIR_NETWORK_FORWARD_NAT:
    case VIR_NETWORK_FORWARD_ROUTE:
4730
        ipdef = virNetworkDefGetIPByIndex(netdef, AF_UNSPEC, 0);
4731
        if (!ipdef) {
4732
            virReportError(VIR_ERR_INTERNAL_ERROR,
4733
                           _("network '%s' doesn't have an IP address"),
4734
                           netdef->name);
4735
            goto cleanup;
4736 4737 4738 4739 4740
        }
        addrptr = &ipdef->address;
        break;

    case VIR_NETWORK_FORWARD_BRIDGE:
4741
        if ((dev_name = netdef->bridge))
4742 4743
            break;
        /*
4744 4745
         * fall through if netdef->bridge wasn't set, since that is
         * macvtap bridge mode network.
4746 4747 4748 4749
         */
    case VIR_NETWORK_FORWARD_PRIVATE:
    case VIR_NETWORK_FORWARD_VEPA:
    case VIR_NETWORK_FORWARD_PASSTHROUGH:
4750 4751
        if ((netdef->forward.nifs > 0) && netdef->forward.ifs)
            dev_name = netdef->forward.ifs[0].device.dev;
4752

4753
        if (!dev_name) {
4754 4755 4756
            virReportError(VIR_ERR_INTERNAL_ERROR,
                           _("network '%s' has no associated interface or bridge"),
                           netdef->name);
4757
            goto cleanup;
4758 4759 4760 4761
        }
        break;
    }

4762
    if (dev_name) {
4763
        if (virNetDevGetIPAddress(dev_name, &addr) < 0)
4764
            goto cleanup;
4765
        addrptr = &addr;
4766 4767
    }

4768 4769
    if (!(addrptr &&
          (*netaddr = virSocketAddrFormat(addrptr)))) {
4770
        goto cleanup;
4771 4772
    }

4773
    ret = 0;
4774
 cleanup:
4775
    virNetworkObjEndAPI(&network);
4776 4777
    return ret;
}
4778

4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841 4842 4843 4844 4845 4846 4847 4848 4849 4850
/* networkGetActualType:
 * @dom: domain definition that @iface belongs to
 * @iface: the original NetDef from the domain
 *
 * Looks up the network reference by iface, and returns the actual
 * type of the connection without allocating any resources.
 *
 * Returns 0 on success, -1 on failure.
 */
int
networkGetActualType(virDomainNetDefPtr iface)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
    virNetworkObjPtr network = NULL;
    virNetworkDefPtr netdef = NULL;
    int ret = -1;

    if (!driver || iface->type != VIR_DOMAIN_NET_TYPE_NETWORK)
        return iface->type;

    if (iface->data.network.actual)
        return iface->data.network.actual->type;

    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!network) {
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
        return -1;
    }
    netdef = network->def;

    if ((netdef->forward.type == VIR_NETWORK_FORWARD_NONE) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_NAT) ||
        (netdef->forward.type == VIR_NETWORK_FORWARD_ROUTE)) {
        /* for these forward types, the actual net type really *is*
         * NETWORK; we just keep the info from the portgroup in
         * iface->data.network.actual
         */
        ret = VIR_DOMAIN_NET_TYPE_NETWORK;

    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) &&
               netdef->bridge) {

        /* <forward type='bridge'/> <bridge name='xxx'/>
         * is VIR_DOMAIN_NET_TYPE_BRIDGE
         */

        ret = VIR_DOMAIN_NET_TYPE_BRIDGE;

    } else if (netdef->forward.type == VIR_NETWORK_FORWARD_HOSTDEV) {

        ret = VIR_DOMAIN_NET_TYPE_HOSTDEV;

    } else if ((netdef->forward.type == VIR_NETWORK_FORWARD_BRIDGE) ||
               (netdef->forward.type == VIR_NETWORK_FORWARD_PRIVATE) ||
               (netdef->forward.type == VIR_NETWORK_FORWARD_VEPA) ||
               (netdef->forward.type == VIR_NETWORK_FORWARD_PASSTHROUGH)) {

        /* <forward type='bridge|private|vepa|passthrough'> are all
         * VIR_DOMAIN_NET_TYPE_DIRECT.
         */

        ret = VIR_DOMAIN_NET_TYPE_DIRECT;

    }

    virNetworkObjEndAPI(&network);
    return ret;
}


4851 4852 4853
/**
 * networkCheckBandwidth:
 * @net: network QoS
4854
 * @ifaceBand: interface QoS (may be NULL if no QoS)
4855
 * @oldBandwidth: new interface QoS (may be NULL if no QoS)
4856
 * @ifaceMac: interface MAC (used in error messages for identification)
4857 4858
 * @new_rate: new rate for non guaranteed class
 *
4859 4860 4861 4862 4863 4864 4865 4866
 * Function checks if @ifaceBand can be satisfied on @net. However, sometimes it
 * may happen that the interface that @ifaceBand corresponds to is already
 * plugged into the @net and the bandwidth is to be updated. In that case we
 * need to check if new bandwidth can be satisfied. If that's the case
 * @ifaceBand should point to new bandwidth settings and @oldBandwidth to
 * current ones. If you want to suppress this functionality just pass
 * @oldBandwidth == NULL.
 *
4867 4868 4869 4870 4871 4872
 * Returns: -1 if plugging would overcommit network QoS
 *           0 if plugging is safe (@new_rate updated)
 *           1 if no QoS is set (@new_rate untouched)
 */
static int
networkCheckBandwidth(virNetworkObjPtr net,
4873
                      virNetDevBandwidthPtr ifaceBand,
4874
                      virNetDevBandwidthPtr oldBandwidth,
4875
                      virMacAddr ifaceMac,
4876 4877 4878 4879 4880 4881 4882 4883
                      unsigned long long *new_rate)
{
    int ret = -1;
    virNetDevBandwidthPtr netBand = net->def->bandwidth;
    unsigned long long tmp_floor_sum = net->floor_sum;
    unsigned long long tmp_new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];

4884
    virMacAddrFormat(&ifaceMac, ifmac);
4885 4886 4887 4888 4889 4890 4891 4892 4893 4894

    if (ifaceBand && ifaceBand->in && ifaceBand->in->floor &&
        !(netBand && netBand->in)) {
        virReportError(VIR_ERR_OPERATION_UNSUPPORTED,
                       _("Invalid use of 'floor' on interface with MAC "
                         "address %s - network '%s' has no inbound QoS set"),
                       ifmac, net->def->name);
        return -1;
    }

4895 4896
    if (((!ifaceBand || !ifaceBand->in || !ifaceBand->in->floor) &&
         (!oldBandwidth || !oldBandwidth->in || !oldBandwidth->in->floor)) ||
4897 4898
        !netBand || !netBand->in) {
        /* no QoS required, claim success */
4899
        return 1;
4900
    }
4901 4902

    tmp_new_rate = netBand->in->average;
4903 4904 4905 4906
    if (oldBandwidth && oldBandwidth->in)
        tmp_floor_sum -= oldBandwidth->in->floor;
    if (ifaceBand && ifaceBand->in)
        tmp_floor_sum += ifaceBand->in->floor;
4907 4908 4909 4910 4911 4912 4913 4914 4915 4916 4917 4918 4919 4920 4921 4922 4923 4924 4925 4926 4927 4928 4929 4930 4931

    /* check against peak */
    if (netBand->in->peak) {
        tmp_new_rate = netBand->in->peak;
        if (tmp_floor_sum > netBand->in->peak) {
            virReportError(VIR_ERR_OPERATION_INVALID,
                           _("Cannot plug '%s' interface into '%s' because it "
                             "would overcommit 'peak' on network '%s'"),
                           ifmac,
                           net->def->bridge,
                           net->def->name);
            goto cleanup;
        }
    } else if (tmp_floor_sum > netBand->in->average) {
        /* tmp_floor_sum can be between 'average' and 'peak' iff 'peak' is set.
         * Otherwise, tmp_floor_sum must be below 'average'. */
        virReportError(VIR_ERR_OPERATION_INVALID,
                       _("Cannot plug '%s' interface into '%s' because it "
                         "would overcommit 'average' on network '%s'"),
                       ifmac,
                       net->def->bridge,
                       net->def->name);
        goto cleanup;
    }

4932 4933
    if (new_rate)
        *new_rate = tmp_new_rate;
4934 4935
    ret = 0;

4936
 cleanup:
4937 4938 4939 4940 4941 4942 4943 4944 4945 4946 4947 4948 4949 4950 4951 4952
    return ret;
}

/**
 * networkNextClassID:
 * @net: network object
 *
 * Find next free class ID. @net is supposed
 * to be locked already. If there is a free ID,
 * it is marked as used and returned.
 *
 * Returns next free class ID or -1 if none is available.
 */
static ssize_t
networkNextClassID(virNetworkObjPtr net)
{
4953
    ssize_t ret = 0;
4954

4955
    ret = virBitmapNextClearBit(net->class_id, -1);
4956

4957
    if (ret < 0 || virBitmapSetBit(net->class_id, ret) < 0)
4958 4959 4960 4961 4962
        return -1;

    return ret;
}

4963

4964
static int
4965 4966 4967 4968
networkPlugBandwidthImpl(virNetworkObjPtr net,
                         virDomainNetDefPtr iface,
                         virNetDevBandwidthPtr ifaceBand,
                         unsigned long long new_rate)
4969
{
4970
    virNetworkDriverStatePtr driver = networkGetDriver();
4971
    ssize_t class_id = 0;
4972 4973
    int plug_ret;
    int ret = -1;
4974 4975 4976 4977 4978 4979 4980 4981

    /* generate new class_id */
    if ((class_id = networkNextClassID(net)) < 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("Could not generate next class ID"));
        goto cleanup;
    }

4982 4983
    plug_ret = virNetDevBandwidthPlug(net->def->bridge, net->def->bandwidth,
                                      &iface->mac, ifaceBand, class_id);
4984 4985 4986 4987 4988 4989 4990 4991
    if (plug_ret < 0) {
        ignore_value(virNetDevBandwidthUnplug(net->def->bridge, class_id));
        goto cleanup;
    }

    /* QoS was set, generate new class ID */
    iface->data.network.actual->class_id = class_id;
    /* update sum of 'floor'-s of attached NICs */
4992
    net->floor_sum += ifaceBand->in->floor;
4993
    /* update status file */
4994
    if (virNetworkSaveStatus(driver->stateDir, net) < 0) {
4995
        ignore_value(virBitmapClearBit(net->class_id, class_id));
4996
        net->floor_sum -= ifaceBand->in->floor;
4997 4998 4999 5000
        iface->data.network.actual->class_id = 0;
        ignore_value(virNetDevBandwidthUnplug(net->def->bridge, class_id));
        goto cleanup;
    }
5001 5002
    /* update rate for non guaranteed NICs */
    new_rate -= net->floor_sum;
5003
    if (virNetDevBandwidthUpdateRate(net->def->bridge, 2,
5004 5005 5006 5007 5008
                                     net->def->bandwidth, new_rate) < 0)
        VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
                 net->def->bridge);

    ret = 0;
5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048
 cleanup:
    return ret;
}


static int
networkPlugBandwidth(virNetworkObjPtr net,
                     virDomainNetDefPtr iface)
{
    int ret = -1;
    int plug_ret;
    unsigned long long new_rate = 0;
    char ifmac[VIR_MAC_STRING_BUFLEN];
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);

    if ((plug_ret = networkCheckBandwidth(net, ifaceBand, NULL,
                                          iface->mac, &new_rate)) < 0) {
        /* helper reported error */
        goto cleanup;
    }

    if (plug_ret > 0) {
        /* no QoS needs to be set; claim success */
        ret = 0;
        goto cleanup;
    }

    virMacAddrFormat(&iface->mac, ifmac);
    if (iface->type != VIR_DOMAIN_NET_TYPE_NETWORK ||
        !iface->data.network.actual) {
        virReportError(VIR_ERR_INTERNAL_ERROR,
                       _("Cannot set bandwidth on interface '%s' of type %d"),
                       ifmac, iface->type);
        goto cleanup;
    }

    if (networkPlugBandwidthImpl(net, iface, ifaceBand, new_rate) < 0)
        goto cleanup;

    ret = 0;
5049

5050
 cleanup:
5051 5052 5053 5054 5055 5056 5057
    return ret;
}

static int
networkUnplugBandwidth(virNetworkObjPtr net,
                       virDomainNetDefPtr iface)
{
5058
    virNetworkDriverStatePtr driver = networkGetDriver();
5059 5060
    int ret = 0;
    unsigned long long new_rate;
5061
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
5062 5063 5064

    if (iface->data.network.actual &&
        iface->data.network.actual->class_id) {
5065 5066 5067 5068 5069
        if (!net->def->bandwidth || !net->def->bandwidth->in) {
            VIR_WARN("Network %s has no bandwidth but unplug requested",
                     net->def->name);
            goto cleanup;
        }
5070 5071 5072 5073 5074 5075 5076 5077 5078 5079 5080
        /* we must remove class from bridge */
        new_rate = net->def->bandwidth->in->average;

        if (net->def->bandwidth->in->peak > 0)
            new_rate = net->def->bandwidth->in->peak;

        ret = virNetDevBandwidthUnplug(net->def->bridge,
                                       iface->data.network.actual->class_id);
        if (ret < 0)
            goto cleanup;
        /* update sum of 'floor'-s of attached NICs */
5081
        net->floor_sum -= ifaceBand->in->floor;
5082 5083 5084 5085
        /* return class ID */
        ignore_value(virBitmapClearBit(net->class_id,
                                       iface->data.network.actual->class_id));
        /* update status file */
5086
        if (virNetworkSaveStatus(driver->stateDir, net) < 0) {
5087
            net->floor_sum += ifaceBand->in->floor;
5088 5089 5090 5091
            ignore_value(virBitmapSetBit(net->class_id,
                                         iface->data.network.actual->class_id));
            goto cleanup;
        }
5092 5093
        /* update rate for non guaranteed NICs */
        new_rate -= net->floor_sum;
5094
        if (virNetDevBandwidthUpdateRate(net->def->bridge, 2,
5095 5096 5097 5098 5099 5100 5101
                                         net->def->bandwidth, new_rate) < 0)
            VIR_WARN("Unable to update rate for 1:2 class on %s bridge",
                     net->def->bridge);
        /* no class is associated any longer */
        iface->data.network.actual->class_id = 0;
    }

5102
 cleanup:
5103 5104
    return ret;
}
5105 5106 5107

static void
networkNetworkObjTaint(virNetworkObjPtr net,
5108
                       virNetworkTaintFlags taint)
5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119
{
    if (virNetworkObjTaint(net, taint)) {
        char uuidstr[VIR_UUID_STRING_BUFLEN];
        virUUIDFormat(net->def->uuid, uuidstr);

        VIR_WARN("Network name='%s' uuid=%s is tainted: %s",
                 net->def->name,
                 uuidstr,
                 virNetworkTaintTypeToString(taint));
    }
}
5120 5121 5122 5123 5124 5125


static bool
networkBandwidthGenericChecks(virDomainNetDefPtr iface,
                              virNetDevBandwidthPtr newBandwidth)
{
5126
    virNetDevBandwidthPtr ifaceBand;
5127 5128 5129 5130 5131 5132 5133 5134
    unsigned long long old_floor, new_floor;

    if (virDomainNetGetActualType(iface) != VIR_DOMAIN_NET_TYPE_NETWORK) {
        /* This is not an interface that's plugged into a network.
         * We don't care. Thus from our POV bandwidth change is allowed. */
        return false;
    }

5135
    ifaceBand = virDomainNetGetActualBandwidth(iface);
5136 5137 5138 5139 5140 5141 5142 5143 5144 5145 5146 5147 5148 5149 5150 5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169 5170 5171 5172 5173 5174 5175
    old_floor = new_floor = 0;

    if (ifaceBand && ifaceBand->in)
        old_floor = ifaceBand->in->floor;
    if (newBandwidth && newBandwidth->in)
        new_floor = newBandwidth->in->floor;

    return new_floor != old_floor;
}


bool
networkBandwidthChangeAllowed(virDomainNetDefPtr iface,
                              virNetDevBandwidthPtr newBandwidth)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
    virNetworkObjPtr network = NULL;
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
    bool ret = false;

    if (!networkBandwidthGenericChecks(iface, newBandwidth))
        return true;

    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!network) {
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
        return false;
    }

    if (networkCheckBandwidth(network, newBandwidth, ifaceBand, iface->mac, NULL) < 0)
        goto cleanup;

    ret = true;

 cleanup:
    virNetworkObjEndAPI(&network);
    return ret;
}
5176 5177 5178 5179 5180 5181 5182 5183 5184 5185 5186 5187 5188 5189 5190 5191 5192 5193 5194 5195 5196 5197 5198 5199 5200 5201 5202 5203 5204 5205 5206 5207 5208 5209 5210 5211 5212 5213


int
networkBandwidthUpdate(virDomainNetDefPtr iface,
                       virNetDevBandwidthPtr newBandwidth)
{
    virNetworkDriverStatePtr driver = networkGetDriver();
    virNetworkObjPtr network = NULL;
    virNetDevBandwidthPtr ifaceBand = virDomainNetGetActualBandwidth(iface);
    unsigned long long new_rate = 0;
    int plug_ret;
    int ret = -1;

    if (!networkBandwidthGenericChecks(iface, newBandwidth))
        return 0;

    network = virNetworkObjFindByName(driver->networks, iface->data.network.name);
    if (!network) {
        virReportError(VIR_ERR_NO_NETWORK,
                       _("no network with matching name '%s'"),
                       iface->data.network.name);
        return ret;
    }

    if ((plug_ret = networkCheckBandwidth(network, newBandwidth, ifaceBand,
                                          iface->mac, &new_rate)) < 0) {
        /* helper reported error */
        goto cleanup;
    }

    if (plug_ret > 0) {
        /* no QoS needs to be set; claim success */
        ret = 0;
        goto cleanup;
    }

    /* Okay, there are three possible scenarios: */

5214 5215
    if (ifaceBand && ifaceBand->in && ifaceBand->in->floor &&
        newBandwidth && newBandwidth->in && newBandwidth->in->floor) {
5216 5217 5218 5219 5220 5221 5222 5223 5224 5225 5226 5227 5228 5229 5230 5231 5232 5233 5234 5235 5236 5237 5238 5239 5240 5241 5242 5243 5244 5245 5246 5247 5248 5249 5250 5251 5252 5253 5254 5255 5256 5257
        /* Either we just need to update @floor .. */

        if (virNetDevBandwidthUpdateRate(network->def->bridge,
                                         iface->data.network.actual->class_id,
                                         network->def->bandwidth,
                                         newBandwidth->in->floor) < 0)
            goto cleanup;

        network->floor_sum -= ifaceBand->in->floor;
        network->floor_sum += newBandwidth->in->floor;
        new_rate -= network->floor_sum;

        if (virNetDevBandwidthUpdateRate(network->def->bridge, 2,
                                         network->def->bandwidth, new_rate) < 0 ||
            virNetworkSaveStatus(driver->stateDir, network) < 0) {
            /* Ouch, rollback */
            network->floor_sum -= newBandwidth->in->floor;
            network->floor_sum += ifaceBand->in->floor;

            ignore_value(virNetDevBandwidthUpdateRate(network->def->bridge,
                                                      iface->data.network.actual->class_id,
                                                      network->def->bandwidth,
                                                      ifaceBand->in->floor));
            goto cleanup;
        }
    } else if (newBandwidth->in && newBandwidth->in->floor) {
        /* .. or we need to plug in new .. */

        if (networkPlugBandwidthImpl(network, iface, newBandwidth, new_rate) < 0)
            goto cleanup;
    } else {
        /* .. or unplug old. */

        if (networkUnplugBandwidth(network, iface) < 0)
            goto cleanup;
    }

    ret = 0;
 cleanup:
    virNetworkObjEndAPI(&network);
    return ret;
}